The BioMeter: An in vivo tissue biosensor for fertility awareness and possible additional applications.

Part 2. The BioMeter and how it operates.


<>Introduction

The BioMeter is an electronic tool for personal fertility awareness and natural family planning, with potential to serve as a broadly applicable aid in women's healthcare. It should be useful in the management of impaired fertility. Examples of additional prospective applications are early pregnancy detection, birth-date pre-determination, the management of premenstrual syndrome, the timing of breast self-examinations, management of hormone replacement therapy, timing of medication intake in general - including certain new types of oral contraceptives. Further potential applications are prospectively as a parturition alarm, and speculatively as a pre-screen for early warning of cervical cancer development and of pelvic inflammatory disease (salpingitis). The BioMeter could also be useful in the determination of bio-equivalence of certain steroid compounds.

The BioMeter monitor is an in vivo tissue-assay biosensor. In a normal healthy human female, it provides two predictive signals that in principle should permit the determination of the beginning of the fertile period, and another signal marking ovulation. The fertility markers are believed to be associated with the stages of folliculogenesis. It is believed that the technology thus offers a definitive solution to the problem of determining the fertility window, a problem that has eluded solution ever since the existence of the brief periodic recurrence of fertility in the female has been recognized.

As such the BioMeter technology has potential to add significant value to the market of personal diagnostic aids. This is based on the principle of the method and on the nature of the results that it has generated in the hands of independent investigators who tested prototypes of the BioMeter in three different species of mammalian females, namely bovine and porcine experimental subjects in addition to human volunteers. The results of these experiments are internally consistent - in terms of consistent response to the sex steroids - while exhibiting species differences such as the follicular waves in the animal data versus folliculogenesis of the single dominant follicle in the human female.

The potential usefulness of the BioMeter stems from the fact that it generates, in a quick and easy manner, a menstrual cyclic pattern that has a high information content by virtue of a number of consistently occurring peaks, and a separate marker of ovulation. The cyclic profile data can be processed electronically so as to provide interpreted diagnostic information, obviating the need for the user to work with raw data.

In addition to several peaks in the post-ovulation (luteal) phase of the profile, tentatively associated with the pulsatile release of progesterone from the corpus luteum and with the elevated estrogen, there are the two essential signals in the pre-ovulation (follicular) phase of the cyclic profile. These are believed to be driven by the ovarian steroid hormones via local pathways, not merely via general circulation. Any delay between the hormonal input and the actual occurrence of ovulation, such as in asynchronous cycles, is detected by the BioMeter as a matter of course.

The BioMeter can do this because it detects the reproductive status by means of electronic coupling with certain tissues of the reproductive tract. The tissues are uniquely sensitive to the regulatory signals from the ovary, the timing regulator of the menstrual cycle [2]. Unlike the older methods of fertility monitoring, the BioMeter has a marker of ovulation that is distinct from the predictive signals, detecting ovulation independently of the predictive markers, through a tightly coupled mechanism. Moreover, the BioMeter works in a convenient, elegant and user-friendly manner, at a very affordable cost. These features are also potentially an important added value to the healthcare market.

The BioMeter has resulted from the initial efforts of an apparently infertile married couple of scientists, one working in gynecology and the other in physical chemistry/electrochemistry with focus on biomedical problems. As a result of this project, it became clear that the once-perceived infertility problem was a misconception and that, in fact, the problem was one of sub-fertility (or reduced fertility, defined as the inability to conceive from less than a year of repeated unprotected intercourse). In this case, proper timing of the attempt to conceive did turn out to resolve the problem, on two occasions. The BioMeter development project has been a privately funded effort, within an entrepreneurial endeavor to bring about a company specialized in electronic fertility diagnosis.



The BioMeter method


In contrast to the Zetek Cue instrument, the BioMeter involves only the one vaginal probe with which the "smart" electronics are integrated into a small, unthreatening and, in fact, feminine-looking package. This design feature, along with the safety of operation and simplicity of use, is considered an essential attribute for a device offered as a daily companion tool for women's healthcare.

The design of the BioMeter stems from several guiding points or principles. The electrometric procedure is focused on the cervix uteri as the biological valve in which the fertility status resides in the best-defined, most specific and accessible, manner. Ovarian vein-to-artery exchange of steroids, prostaglandins and other bioactive substances is a local transfer mechanism. As such, it enables local regulation of the genital organs [3]. In addition, the cervix, like the isthmus region of the fallopian tubes, has a particularly dense innervation, specifically by the vasoactive intestinal polypeptide nerve [4]. Potentiometric detection of menstrual cyclic phenomena has been known since the classical work of Burr at Yale in the thirties, but we decided to avoid the use of any reference electrode, and to tap into the detectable periodicity by other means. Admittance monitoring was selected and modified so as to achieve safety of operation and simplicity of design.

Certain concepts and various studies have provided stimulating pointers for our approach but any such stimuli were modulated by practical considerations, with the intended non-technical user in mind.

The optimal parameters for the design of the hand-held model were established empirically, which is also how we decided on the tissue contact in the posterior fornix region. Many combinations of frequency, inter-electrode potential difference, and electrode size and positioning were examined before an optimal combination was selected for the miniaturized device. This is described as applying a potential difference so small that no electrochemical reaction can occur (an order of magnitude below such electrolysis-prone levels), at a sufficiently high frequency, using fairly small surface area electrodes that allow contact within the region of posterior fornix in a reproducible electrode orientation. In this manner, an innocuous capacitive coupling of the electronics with the target epithelium yields a tissue biosensor set up momentarily for the measurement.



Procedure


Each subject was provided with a prototype unit for her individual use at home. In the instructions for use, the woman-user was advised to apply the probe as she would a vaginal tampon, but of course only for as long as it takes to obtain a reading (a few seconds). We have not dictated any particular position for the insertion, leaving it to the woman to decide whether to insert while standing up or sitting down, kneeling or lying on her back. Once selected, the mode of insertion should be consistent from day to day. The instructions for use contained the tip that measurements are best done with an empty bladder, and the explanation that, if inserted like a vaginal tampon, as far as it goes, the front end of the sensor would be in the posterior fornix of the vagina. The electronics are programmed to turn power off after 20 seconds, which usually provides enough margin even during the critical days around ovulation when the readings are least stable. A steady-state reading is required, and the measurements are done at about the same time of day (+/- about 1 hour), to avoid the expectable effects of circadian rhythm on the measurements.

In the experiments reported here, the woman-user decided on the steady state by observing the numerical display, but in the planned smart version the decision will be performed electronically, along with other functions. The smart version has not yet been completed although it has been designed. Subjects recorded daily the date, day of cycle, the measurement data, and the data from the reference methods. The reference methods were the BBT in the Torino study, but two other methods were employed in the Marquette study, namely an LH kit indication and a vaginal-cervical mucus self-evaluation according to the Creighton model criteria. In a separate experiment that is not presented here, laboratory-performed analyses of morning urine were performed to correlate the monitor data with urinary LH and FSH (see Results).



Subjects


All data presented here are from spontaneously ovulating subjects. An early prototype of the BioMeter was tested by a gynecologist without any vested interest in the technology, at the First Gynecology Clinic, University of Turin, Italy. The gynecologist selected four healthy nulliparous volunteers of 25 to 30 years of age, who were not using any medication or birth control, so as to generate baseline pattern data for the monitor. We shall now use these baseline results to introduce the monitor, and then compare the baseline profiles with other results that have been generated by non-baseline subjects. The presentation of the non-baseline results will include also those from a more recent study of another prototype of the technology. This was performed at a natural family planning (NFP) clinic at Marquette University in Milwaukee, Wisconsin. In the Marquette study, 21 cycles were monitored by 10 non-baseline women. These subjects ranged in age from 19 to 42 years, and were both nulliparous and multiparous. These women and their cycles therefore represented a broad cross-section of the population of reproductive-age women. They included both early and older reproductive ages, and both those women who use NFP for birth control as well as those attending the clinic to utilize NFP methodology as a conception aid.



Results


Figure 1 introduces the BioMeter menstrual cyclic profile, using data generated by a baseline-type volunteer subject carefully selected by the investigating gynecologist in the first pilot study. The cyclic pattern exhibits a number of well-defined peaks and troughs, appearing in both the morning graph (AM2) and the evening graph (PM2) in Figure 1. The same features appear also in the cyclic profiles obtained in other cycles by other subjects. The data point on day 6 is the first post-menstruation minimum or trough, which occurs typically on day 6, 7 or 8. This timing corresponds to the timing of the selection of the dominant follicle as known in reproductive endocrinology. The readings then rise to a peak on day 9, the highest reading of the cycle. This is the long-term predictive peak, which occurs at a time that varies from cycle to cycle, sometimes about a week before and in other cycles much closer to the yet to be introduced ovulation marker.

The acute warning of the impending ovulation in the profile of Figure 1 is provided by the peak on day 15, called the short-term predictive peak. This seems to have the timing of a signal from the ovulating ovary that ovulation of the selected follicle is imminent. The sensor reading then falls off directly into the trough on day 16, which has been interpreted as the marker of ovulation. This is the lowest reading of the cycle, always around the same signal amplitude, which is near 100 microamp units at the given calibration. The corresponding basal body temperature (BBT) curve rises to the post-ovulation elevated level on the following day, day 17 of this cycle. Such a BBT rise after the marker minimum has been observed in all cycles where the BBT was taken alongside the BioMeter measurements.

The "mid-cycle" minimum in the cyclic pattern has also been found to coincide with the day of the luteinizing hormone (LH) and follicle stimulating hormone (FSH) peaks in the woman's urine, and it is therefore considered to be the instrument's ovulation marker. This correlation was determined in a blind manner by a Serono Diagnostics laboratory using monthly series of frozen samples of morning urine. Because the concentrations of the two pituitary peptide hormones are known to peak in the urine within some 3 to 12 hours before ovulation, the correlation was a good indication that the "mid-cycle" minimum coincides with the day of ovulation. The concurrently generated BBT profiles, always showing the post-ovulation temperature rise after the ovulation marker day, have been consistent with the conclusion. This interpretation of the "mid-cycle" minimum is yet to be confirmed by correlation with ultrasound visualization of the rupture of the follicle.

On the days after the ovulation-marker minimum, the BioMeter readings rise to go through several peaks and troughs but do not reach the high levels observed in the pre-ovulation phase of the menstrual cycle. This repeatable post-ovulation trend is interpreted as the effect of pulsatile release of progesterone in the presence of elevated estrogens during the progesterone-dominated luteal phase.

Figure 2 demonstrates the correlation of the cyclic profiles yielded by three different baseline-type women. They were healthy subjects under 35 years of age who were chemically clean non-smokers using no contraception and no other medication. These women were considered baseline subjects characterized by minimal potential complications of physiological or biochemical nature that could cause deviations from norm (or "baseline").

The three profiles in Figure 2 are superimposed at the day of the "mid-cycle" minimum, which - as already stated - is interpreted as the marker of ovulation. The repeatability of the ovulation marker and of the short-term predictive peak (the acute warning of impending ovulation) is evident in Figure 2, as is the repeatability of the other features of the cyclic profile. The timing of the first predictive peak (the long-term predictive peak) varies with the length of the menstrual cycle. Thus, the first peak anticipates the ovulation marker minimum by 7 days in the 30 days long cycle PM2, by 5 days in the 28 days long cycle PM1, and by 3 days in the 26 days long cycle PM3. Such a relationship, whereby the farther ahead of the ovulation marker is the long-term predictive peak, the longer is the resulting menstrual cycle, makes sense. It corresponds to the fact that, in normal healthy menstrual cycles, the total length of the cycle depends on the extent of the pre-ovulation phase, with the post-ovulation phase remaining constant at about 14 days [5].

Figure 3 shows data from an abnormal cycle (LPD-25 YRS OLD) against a baseline cycle (PM1-26 YRS OLD) for comparison. The abnormal cycle is from one of the baseline-type subjects, and is believed to be a case of the luteal phase defect (LPD), a problem which occurs quite frequently and causes the failure to conceive in otherwise healthy women. The LPD interpretation of the aberrant cyclic data is based on the absence of the predictive peaks in the pre-ovulation phase of the cyclic profile, coupled with the fact that the corresponding basal body temperature curve (BBT LPD) exhibits a corrupted profile rather than the sustained elevated temperature plateau. Such a temperature profile is associated with the luteal phase defect, which is believed to be a consequence of ovarian failure to produce a mature dominant follicle. The BioMeter sensor appears to have detected the failure to produce the dominant follicle through the absence of the two predictive peaks that would otherwise appear in the follicular phase of the cyclic profile. Because of the absence of the follicular peaks, the sensor in effect anticipated the LPD.

We now present data that help to understand the response of the sensor in the course of the reproductive cycle. The following experiment with surgically altered animal females indicates that the steroid hormones have a strong effect on the response of the BioMeter sensor. Figure 4 displays the results of an experiment that are indicative of the opposing effects of the two steroids on the response of the BioMeter sensor: estrogen drives the response in one direction while progesterone pushes the readings in the opposite direction.

The surgically ovariectomized pubertal female pig was deficient in the steroid hormones. After recovery from surgery, orally administered progesterone (progestagen, allyl-trenbolone) caused over time the sensor readings to decrease from the initial amplitude. Upon discontinuation of the progesterone, the readings returned to the initial level at about the same rate. Then, about a hundred times lower doses of estrogen (estradiol benzoate), administered via intra-muscular injections, caused an increase in the readings, which appeared to be higher than the response to the oral progesterone. Upon discontinuation of the injected estrogen, the sensor readings decreased, and the rate of change was again greater than the response to oral progesterone.

These effects of exogeneous steroids are consistent with the features of the reproductive cyclic profile in the three mammalian species studied with the instrument (human, porcine and bovine females). The results of reproductive cycle monitoring are interpreted as effects of endogeneous steroids on the steroid receptors of the tissues in contact with the sensor electrodes. Similarly consistent results were obtained with ovariectomized cows and with a menopausal human subject who was on steroid hormone replacement therapy.

Figure 5 illustrates three BioMeter cyclic profiles that were generated by a non-baseline subject (cycles 4, 5 and 6), and are shown for comparison with the baseline cycles in Figures 1 and 2. Figure 5A displays the three non-baseline cycles superimposed on the ovulation marker minimum, and Figures 5B, 5C and 5D show the individual cyclic profiles alone, along with the concurrently measured BBT data. The non-baseline profiles exhibit the same features as those observed in the baseline profiles of Figure 2. We also note that the range of amplitudes of the readings is the same as the dynamic range of the baseline profiles.

There are, however, certain quantitative differences, particularly in the timing of the various features. Thus, cycle 4 exhibits an atypical second predictive peak (the acute warning of impending ovulation) in that the short-term predictive peak does not fall directly into the ovulation marker minimum: the interval to ovulation is 3 days rather than 1 day long. This delay in the occurrence of the ovulation marker is corroborated by the corresponding BBT pattern, to the extent that the BBT can be relied upon to confirm that ovulation has occurred. The delay in ovulation is understood as a consequence of a lack of synchronization between the ovarian and neuroendocrine components of the reproductive biorhythm system. This interpretation is linked with the other distinct deviation of cycle 4 from norm, namely that the luteal phase of this cycle is only 10 days long. Luteal phase shorter than 11 days is a significant aberrant deviation from the normal duration of the luteal phase.

This short luteal phase deviation is different from the presumed luteal phase defect in Figure 3, and the short luteal phase is also present in the subsequent cycle of the same non-baseline subject, cycle 5. Cycle 5 exhibits again the expected features but the luteal phase is again too short, this time only 8 days long, whereas the follicular phase is of a 15-day duration. As with cycle 4, the BBT temperature profile is consistent with the BioMeter data, showing elevated temperatures after the day of the ovulation marker minimum. Finally in Figure 5, cycle 6 exhibits a normal extent of the luteal phase, after a somewhat high but otherwise normal short-term predictive peak, which appears after a somewhat unusual but recognizable long-term predictive peak.

Thus, we have seen that the non-baseline cyclic profiles appear to be qualitatively similar to the baseline profiles, with some quantitative deviations from the baseline pattern. Figure 5 also illustrates the problem of missing data points, a problem that may be expected to come up particularly in the non-investigational use of the technology. In this example, cycle 4 has some missing readings within the long-term predictive peak but the cyclic pattern is still recognizable; the missing data in this case does not interfere with the interpretation of the rest of the cyclic profile.

A more recent pilot study of a somewhat upgraded prototype was conducted in the setting of a natural family planning (NFP) clinic at Marquette University in Milwaukee, Wisconsin. The prototypes were upgraded in terms of the electronics, which included some more sophisticated functions than the earlier model. The study was supported by the Marquette University Regular Research Grant and designed, conducted and published independently of the present writer [6]. Let us now review the published data, examining certain aspects that were not brought out by the investigators. Analyzing the data in the context of the results described above should provide further insight into the BioMeter technique.

In the Marquette study, 21 cycles were monitored by 10 non-baseline women who ranged in age from 19 to 42 years, and were both nulliparous and multiparous. These women and their cycles therefore represent a broad cross-section of the population of reproductive-age women, including both early and older reproductive ages, and women interested in NFP as birth-control as well those using NFP as a conception aid because of difficulties in conceiving.

The three monitor-generated markers of the fertility status were compared with the data of two reference methods. The reference methods were the LH kit for urine analysis, and the self-rated assessment of the vaginal-cervical mucus as practiced at the Marquette NFP clinic (utilizing the Creighton model criteria), which corresponds to the peak mucus method employed in the study by France et al. as discussed above. The Marquette study did not utilize the ultrasound gold standard.

The Marquette study found that the monitor's ovulation marker, O, and the short-term predictive marker, OS, correlated strongly with the two reference methods. There was a narrow range of the short-term predictive intervals, O - OS, but a wide range of the long-term predictive intervals, O - OL (where OL is the long-term predictive peak). The abbreviations are:
OL --- long-term predictive peak
OS --- short-term predictive peak
O --- ovulation marker.

An examination of the Marquette study data reveals the following two categories of menstrual cycle data as yielded by the non-baseline type of subjects enrolled in the study. We discarded data on cycle 3.1 because they indicated OL=12 and OS=13, an impossible sequence (without a day for a nadir between the peaks).



I. Regular cycles

Table 1 shows that in 11 of the 20 recorded cycles (55%) the ovulation marker, O, falls within one day of either the LH surge or the peak mucus (Pk) or both. This is a good agreement, considering the inherently limited accuracy of the two reference methods, as discussed in Part 1 of this paper.
Please scroll down to Table 1 and to the continued text.

Table 1. REGULAR CYCLES
DEFINED AS THOSE WHERE O IS WITHIN 1 DAY OF LH OR Pk

subject.cycle#	age,parity	OL	OS	O	LH	Pk
1.1	35,M	11	14	16	17	17
1.2	35,M	7	13	17	16	16
2.2	33,M	6	12	14	14	14
7.2	29,N	8	14	15	14	13
8.1	19,N	12	18	21	21	23
9.2	41,N	9	12	13	13	14
9.4	41,N	7	13	15	14	13
10.2	22,N	6	8	10	10	12

We further observe that, in this group of menstrual cycles, the temporal relationship between the OS (short-term predictive peak) and the reference indicator LH is
OS < LH,
i.e., the short-term predictive peak occurs before the urinary LH indication. Since the short-term predictor OS is thought to be due to an ovarian signal indicating the readiness to ovulate, this temporal relationship between the OS and the LH is as would be expected because the urinary LH occurs practically concurrently with ovulation. The OS < LH relationship does not hold for two of the cycles, 7.1 and 7.2, in which OS = LH.




II. Irregular cycles

Table 2 shows that in 9 of the 21 cycles in the Marquette study (43%), the ovulation marker, O, is not within 1 day of either of the reference indicators, and that it always follows later, delayed by 2 to 3 days. In these so-called irregular cycles, the sensor detects ovulation 2 or 3 days later than indicated by the reference indicators LH and Pk.
Please scroll down to Table 2 and the continued text.

Table 2. IRREGULAR CYCLES
DEFINED AS THOSE WHERE O IS NOT WITHIN 1 DAY OF LH OR Pk, AND IS ALWAYS HIGHER - SIGNIFYING DELAY OF OVULATION

subject.cycle#	age,parity	OL	OS	O	LH	Pk
2.1	33,M	12	15	17	14	14
4.1	33,N	7	13	15	13	13
2.1	33,M	12	15	17	14	14
6.1	38,M	8	15	19	15	16
6.2	38,M	8	15	18	15	16
8.2	19,N	14	21	24	none	21
9.1	41,N	8	16	18	14	15
9.1	41,N	8	16	18	14	15
10.1	22,N	7	10	12	none	9

We further observe that, in this group of menstrual cycles, the temporal relationship between the OS (short-term predictive peak) and the reference indicator LH is different from that in the regular cycles, namely
OS > LH or OS = LH,
i.e., the short-term predictive peak occurs either after the LH or coincides with it. In these cycles, not only does the ovulation marker O occur at least 2 or 3 days later than both the LH and peak mucus, the short-term predictive peak also occurs later than in the regular cycles (later with respect to the LH indication). These 9 menstrual cycles may be referred to as irregular or "challenged" menstrual cycles.

Thus, defining two categories of cycles depending on the temporal relationship between O and LH (i.e., between the ovulation marker and the urinary LH indication) shows that the two categories of cycles differ in terms of the timing of the short-term predictive peak, OS.

Only two of the ten women enrolled in the Marquette University pilot study presented more than one regular cycle. The other eight subjects had either cycles of both categories or two irregular cycles.

One subject, a 41-years old nulliparous woman, presented the following sequence of four cycles:
an irregular cycle with ovulation marker day O=18,
followed by a regular cycle with O=13,
followed by an irregular cycle with O=18,
followed by a regular cycle with O=15.

Only two among the regular cycles in Table 1 have numerical attributes that are consistent with the concept of baseline cyclic profile (namely, cycles 7.2 and 9.2). It should be interesting to see to what extent these results may be representative of the general population. To that end, a large-scale investigation of the technique introduced here is needed.



Discussion


The BioMeter is presented as an in vivo monitor of female reproductive physiology, of which the principal component is ovarian function. Ovarian function is a key component in the regulation of the menstrual cycle, and as such is of primary importance in women's healthcare, as Toth and Hodgen put it, "from intrauterine development to the postmenopausal status" [7]. The BioMeter design is also cognizant of the other fundamental principle of reproductive physiology, as expressed in an authoritative citation that emphasizes this aspect: "Ovulation is a spontaneous periodic phenomenon but neural mechanisms are also involved. In women, menstrual cycles may be markedly influenced by emotional stimuli" [8], or as put more broadly by Ferin as cited above, disturbances in the menstrual cycle occur in response to exercise and physical demands, stress and emotional demands, and diet and nutritional demands [5].

Certain concepts and various studies provided stimulating pointers for our approach to the technique design. The fuel cell concept of physiological phenomena was the starting point. The epithelium of the targeted region was envisaged to function basically as a network of microscopic fuel cells. In this network, the hormone-responsive electrodic components are the enzymes that drive the documented cyclical changes in the composition of the epithelial cells and in the mucus secretions of the epithelia. Such enzymes in the vaginal epithelia were described for example by Rosa and Velardo [9], who demonstrated pattern distributions of enzymes and of steroid hormone sensitivities for different vaginal regions. These enzymes could be responsive to the hormones by means of associated hormone-receptor sites. The genital tract is known to be rich in the concentrations of the sex hormone receptors.

Information from specialist publications supports or suggests the rationale for the selection of the cervix uteri and/or of the posterior fornix as the locus of the menstrual cycle sensing mechanism. The microvasculature of the reproductive organs may not warrant its own chapters in medical texts but the expert's description and reasoning are meaningful. Verco argues [3] that ovarian vein to artery exchange of bioactive substances enables the local regulation of uterine, tubal and ovarian function. As a result, the genital organs are exposed to higher than central mixed venous or central arterial hormone concentrations, and they see these effectors in real time. With its dense innervation, shared with some other regions of the reproductive tract, the cervical region is a logical choice of tissue for the monitoring of end-organ effects. The end-organ effects include also those that influence the brain and its regulation of the menstrual cycle (such as stress that may delay ovulation or otherwise interfere).

The concept of baseline profiles, obtained from baseline-type subjects, was deemed necessary in order to obtain a frame of reference. It is pertinent to note that the gynecologist, Dr. Chiara Benedetto, reported the results on the baseline cyclic profiles in the form of raw data. The striking reproducibility of the features surrounding day 0 in the profiles of Figure 2 became apparent only after the data were plotted superimposed as presented in Fig. 2. Against the baseline profiles, the similarities and differences noted in the many non-baseline profiles began to make a better sense. The features of the cyclic profile required interpretation, and the endocrinological information on folliculogenesis appeared consistent with the features of the profile repeatedly seen in the follicular phase.

We believe that the pre-ovulation (follicular) phase of the cyclic profile is a reflection of folliculogenesis, which is an endocrinologically defined sequence of intervals or stages in the development of the dominant follicle that is destined to ovulate. The stages are called recruitment, selection, dominance and ovulation [7].

We know that the data point on day 6 in Figure 1 is the first minimum in the profile even though, in this record, the earlier descending part of the curve is missing because the measurements were not done during the first five days of the cycle, on account of the menstrual flow. We know this from experiments where readings were taken despite the menstrual blood, and it was found that the presence of blood did not interfere with the registration of the early, descending, part of the curve. Also, very short cycles such as cycle PM3 in Figure 2, do exhibit the descending part of the curve even after the cessation of menses, thereby resolving the discussed point as the first minimum in the cyclic profile. This first minimum occurs usually on day 6 +/-1 of the cycle. We believe that this marker corresponds to the selection of the dominant follicle, which is the culmination of a recruitment process that has taken place at the end of the previous nonconception cycle, during the menses [7]. It is not clear why the selection should be expressed as a nadir or minimum in the profile but this does seem to be the first important marker of the cyclic profile.

The long-term predictive peak (on day 9 in Figure 1) is believed to be associated with the maturation of the dominant follicle during the interval called dominance. Dominance is characterized by increasing quantities of estrogens, which play an important role in coordinating different parts of the reproductive tract. It is known that the hypothalamic-pituitary axis requires estrogen priming (a certain concentration for at least 36 hours according to [7], citing Knobil). This is a prerequisite for the discharge of an LH surge that would be sufficient for ovulation. Similarly, estrogen stimulation is also required for the cervical epithelium and mucus so as to allow gamete transport at the time of ovulation, as well as for the fallopian tube's embryo transport.

The timing of the long-term predictive peak is found to vary with the length of the menstrual cycle, in baseline and non-baseline cycles with normal luteal phase of about 14 days. Such a relationship, whereby the farther ahead of the ovulation marker is the long-term predictive peak, the longer is the resulting menstrual cycle, corresponds to the fact that, in normal healthy menstrual cycles, the total length of the cycle depends on the extent of the pre-ovulation phase. It is intuitively acceptable that the rate of maturation of the dominant follicle, as reflected in the long-term predictive peak, should control the timing of ovulation and the length of the menstrual cycle. In fact, Eli Adashi has stated this explicitly [10]: "The dominant follicle thus determines the length of the follicular phase, the corpus luteum determining the length of the luteal phase."

The data on the presumed luteal phase defect (LPD), as encountered in one of the baseline-type subjects and seen in Fig. 3, are consistent with the interpretation of the long-term predictive peak in terms of dominant follicle maturation. The absence of the peaks in the follicular phase of the profile is consistent with the failure to produce a dominant follicle, which is believed to be the cause of the LPD. The corrupted shape of the BBT curve of the subject is consistent with the LPD interpretation but, of course, this should be confirmed by additional evidence in further investigations. The luteal phase of the profile will come into focus, examining the hypothesis that the oscillations in that part of the profile are due to pulsatile progesterone release in this progesterone-dominated phase of the cycle. We believe that the range of the readings in the luteal-phase is lower than the range in the follicular phase because of the change from estrogen-dominated to progesterone-dominated regime.

We know that, once the dominant follicle has achieved the necessary size and adequate systemic hormonal effects, final maturational changes within the follicle stimulate ovulation. However, the means by which the oocyte communicates that it is ready for ovulation are not understood. It is not clear whether the increased progesterone in the venous effluent of the ovulating ovary, which occurs apparently at or towards the end of dominance and certainly before ovulation ([11] p.207), may be associated with the appearance of the second minimum in the cyclic profile. The second minimum is the nadir that separates the long-term and the short-term predictive peaks. From the timing of the short-term predictive peak, which occurs before the urinary LH surge (and well before the BBT rise), it seems reasonable to consider the short-term predictive peak as a marker associated with the ovarian signal of the readiness for ovulation.

The final marker of fertility status is the ovulation marker minimum in the cyclic profile. Endocrinologically, the most prominent marker of impending ovulation is the LH surge, which anticipates ovulation within 9 to 12 hours and which is under the control of an ovarian pacemaker that dictates the timing of these events. It is known that at the time of the LH surge, the granulosa cells surrounding the follicle become transformed or luteinized, which means among other things that they become specialized toward synthesis and secretion of progesterone. This causes a rapid increase in progesterone levels that induces a number of changes in the reproductive system. We believe that among them is the effect on the tissues that control the response of the BioMeter sensor, which leads to a tight coupling between ovulation and the sensor.

This notion of tight coupling between ovulation and the tissue-biosensor can perhaps be appreciated on two levels. One, there is a documented plasma concentration rise of progesterone coincident with the LH surge, which includes a large increase of 17-hydroxy-progesterone, about two or three orders of magnitude higher than the levels of estradiol circulating at that time (Ganong [8] page 404 and Speroff et al. [11] page 191). And two, the argument was introduced in the Method section that ovarian vein to artery exchange of bioactive substances enables the local regulation of the genital organs. As a result, these organs are exposed to higher than central mixed venous or central arterial hormone concentrations, and they see these effectors in real time. Christopher Verco goes as far as to say that "such exchange mechanism renders quantification of ovarian steroids in peripheral blood interesting but of little value in predicting the genital end-organ effect" [3].

As can be seen in Fig. 4, progesterone pushes the response of the sensor down, in the opposite direction from the effect of estrogen. This is evident in the data, despite the difficulties encountered in the experiment, including the two different methods of administration of the steroids, which must be reflected in the apparent rates of response. Nevertheless, the overall trends are clear. It is therefore reasonable to consider the "mid-cycle" minimum, the lowest minimum in the profile, to be due to the large burst of progesterone at ovulation. This was validated by the correlation with the urinary LH and FSH peaks, as well as by many concurrently generated BBT profiles, always showing the post-ovulation temperature rise after the ovulation marker day.

The definitive proof is yet to be provided by correlation with ultrasound visualization of the rupture of the follicle [12]. The ultrasound data must be backed by other evidence consistent with ovulation, so as to guard against the possibility of misinterpretation in case of unruptured luteinized follicles or of follicle rupture without egg release. Distinguishing these cases in terms of the BioMeter signature may well contribute to a better understanding of the biosensor, which might in turn contribute to progress in reproductive physiology.

Until the definitive correlation with ultrasound data is generated, we consider the evidence from non-baseline subjects, such as shown in Fig. 5, consistent with the interpretation of the "mid-cycle" minimum as a correlate of ovulation. This includes the atypical cycle 4, in which the short-term predictive peak does not fall directly into the ovulation marker minimum: the interval to the putative ovulation marker is 3 days rather than 1 day long. This delay in the occurrence of the ovulation marker is corroborated by the corresponding BBT pattern (see Fig. 5B), to the extent that the BBT can be relied upon to confirm that ovulation has occurred. The delay in ovulation is understood as a consequence of a lack of synchronization between the ovarian and neuroendocrine components of the reproductive biorhythm system. This interpretation is linked with the other distinct deviation of cycle 4 from norm, namely that the luteal phase of this cycle is only 10 days long. Luteal phase shorter than 11 days is a significant aberrant deviation from the normal duration of the luteal phase.

It will be interesting to find out if the varying amplitudes of the predictive peaks in such non-baseline cycles have a physiological significance. In this context, we note that Figure 2 may suggest, for baseline cycles, a dependence of cycle length not only on the timing of the long-term predictive peak but also on the amplitude of the peak. This, if supported by further data, may suggest the question whether there is a pattern and some physiological significance to the relative amplitudes of the two follicular phase peaks. If so, this might be relevant with respect to the non-baseline and particularly aberrant cycles, and could be useful in the management of reduced fertility or infertility.

The deviations from baseline patterns, when fully understood, may prove useful for the clinicians for whom the details of the cyclic profile should contain meaningful indicators, as is the case with hormonal, BBT and other cyclic patterns. The deviations from baseline do not interfere with the interpretation of the non-baseline cycles in so far as the given deviation is not too drastic as to cause a pathological symptom such as the LPD.

Cycle 4 is a short cycle, as are cycles PM3 and 5, all <28 days. However, unlike cycle PM3, cycles 4 and 5 are abnormal cycles with short luteal phases (<11 days). For short cycles, they both have rather long follicular phases (16 and 15 days, respectively). That is unlike the baseline cycle PM3, which is short simply because of its short follicular phase (with the normal luteal phase of 14 days). The non-baseline cycles 4 and 5 were recorded by a 38-years old woman with a history of ovarian cysts before her two successful pregnancies.

Cycle 4 is even more unusual than cycle 5 because there are three, rather than just one, days of the decreasing readings after the short-term predictive peak. Cycle 4 is tentatively interpreted as a case of asynchrony between follicle maturation and the pituitary signal to ovulate [13]. In such cycles with short luteal phases (<11 days), there is a lack of synchrony due to a mismatch between the ovarian steroids and the pituitary peptides.

The mechanism, which we reference by the phrase "the brain and ovulation" ([11] p. 164) involves the circhoral clock of the hypothalamic GnRH pulse generator, on which the circamensual ovarian clock is "obligatorily dependent" [2]. The GnRH pulse generator is known to be sensitive to stress and to the calming effects of endogeneous enkephalins and beta-endorphin. The opioid tone is an important part of menstrual cyclic function [14]. This is particularly important at the time of the "mid-cycle" LH surge, affecting its timing and intensity [8].

This interpretation of cycle 4 would seem to fit the fact that the short luteal phase is observed more often in women of older reproductive age. Also, the history of the polycystic ovarian syndrome may be pertinent because the syndrome is associated with abnormal luteal phases ([5], p.118).

The data from the Marquette study of non-baseline cycles [6] may fit the concepts under discussion, while the categories of regular and irregular cycles within the non-baseline data are new and their labels tentative. These categories in the limited set of 20 cycles are possibly suggestive of the existence of various categories within the general population. Searching for such patterns should surely be part of future studies. The future study design should take these indications into account.

In the irregular cycles in Table 2, the ovulation marker O occurs later than would be expected from the urinary LH indication, and it is within the three-day uncertainty of the estrogen-driven peak mucus indication. Asynchronous cycles with premature LH surges are a known problem. As many as 35% of menstrual cycles in a normal population have been found abnormal due to asynchrony and a smaller size of the dominant follicle [12] [15]. The number of the irregular cycles is 8% higher in the Marquette study than the 35% of asynchronous cycles reported for normal population (5/14 cycles in [15]). The age and parity composition of this group is such that weighting towards higher percentage of irregularity appears probable: four of the nine cycles are from nulliparous women of 33 and 41 years of age (2 cycles each), two from a 38-years old multiparous, and two are from nulliparous women of 19 and 22 years of age (both did not detect any LH in these irregular cycles and both produced also regular cycles with urinary LH indication).

A frank scrutiny of the results of the Marquette study [6] must bring out certain surprising features in the published data. First, let us state what is not surprising in the data. It is not surprising that the study found a wide range for the timing of the long-term predictive peak, OL (from day 6 to day 14). The authors did not discuss any examination of the relationship between the long-term predictive peak and the length of the menstrual cycles, and they did not consider the reason for the wide range of the O - OL intervals (where O is the ovulation marker). Both a correlation between the OL and cycle length, and a wide range of the O - OL intervals are expected, as discussed above.

It should be pointed out that there is no a priori reason to expect a correlation between OL and O, unlike the expected correlation with the reference methods. (For example, O and LH are expected to correlate because both observables are supposed to mark the same day of ovulation.) This amounts to saying that the algorithm for the use of the OL to determine the width of the fertility window is more complex than a simple proportionality of day numbers or some similar oversimplification.

This brings us to what is surprising in the Marquette study data. It is that most of the cycles, regardless of the category, exhibit O - OS > 1, i.e., the delay-type of the interval between the short-term predictive peak OS and the ovulation marker O (see "the brain and ovulation", above). Although 57% of the cycles fit into the regular category, and although in 54% (i.e., in 6 out of 11) of the regular cycles the ovulation marker O agrees with LH or Pk or both, only two of the so-called regular cycles have O - OS = 1. Thus only two of the cycles, cycles 7.2 and 9.2, would conform to the relationship between the markers that would be expected from the baseline profiles. This statistics, if confirmed, would have serious consequences for the understanding and management of non-baseline cycles in general, and for the design of the smart BioMeter's algorithm in particular.





References


[1] Eli Y. Adashi, John A. Rock, and Zev Rosenwaks, editors, "Reproductive Endocrinology, Surgery, and Technology", Volume 1, Lippincott - Raven, 1996.

[2] J. Hotchkiss and Ernst Knobil, "The hypothalamic pulse generator: The reproductive core", Chapter 7 in [1], pages 124 - 162.

[3] Christopher J. Verco, "Tubal vasculature", Chapter 3, pages 21 - 35, in Alvin M. Siegler, Amir H. Ansari and Leon C. Chesley, editors, The Fallopian Tube. Basic Studies and Clinical Contributions, Futura Publishing Company, 1986.

[4] Christer Owman, Goran Helm, Nils-Otto Sjoberg, and Bengt Walles, "Autonomic neuromuscular mechanisms in the human Fallopian tube", Chapter 4, pages 37 - 53, in Alvin M. Siegler, Amir H. Ansari and Leon C. Chesley, editors, The Fallopian Tube. Basic Studies and Clinical Contributions, Futura Publishing Company, 1986.

[5] Michel J. Ferin, "The menstrual cycle: An integrative view", Chapter 6 in [1], pages 103 - 121.

[6] Richard J. Fehring and William D. Schlaff, "Accuracy of the Ovulon fertility monitor to predict and detect ovulation", Journal of Nurse-Midwifery 43 (No. 2), 117 - 120, 1998.

[7] Thomas L. Toth and Gary D. Hodgen, "Ovarian follicular growth and maturation", Chapter 5, pages 137 -157, in Edward E. Wallach and Howard A. Zacur, editors, Reproductive Medicine and Surgery, Mosby, 1995.

[8] William F. Ganong, Review of Medical Physiology, 17th edition, Chapter 23, "The gonads: Development & function of the reproductive system", Appleton & Lange, 1995.

[9] Charles G. Rosa and J. T. Velardo, "Histochemical localization of vaginal oxidative enzymes and mucins in rats treated with estradiol and progesterone", Annals of New York Academy of Sciences 75 (2), 491 - 503, 1959.

[10] Eli Y. Adashi, "The ovarian follicle: Life cycle of a pelvic clock", Chapter 10 in [1], pages 211 - 234.

[11] Leon Speroff, Robert H. Glass and Nathan G. Kase, "Clinical Gynecologic Endocrinology and Infertility", Williams & Wilkins, 5th edition, 1994.

[12] Frances R. Batzer, "Ultrasonic indices of ovulation", Journal of Reproductive Medicine 31 (No.8), Supplement, 764 - 769, 1986.

[13] A.J. Zeleznik, in E.Y. Adashi and P.K.C. Leung, editors: The Ovary, Raven Press, 1993, pages 41 - 45.

[14] P.R. Gindoff and M. Ferin, Brain opioid peptides and menstrual cyclicity", Seminars in Reproductive Endocrinology 5, 125, 1987. Cited in [11], page 156.

[15] Mary Lake Polan, M. Totora, and B.V. Caldwell, "Abnormal ovarian cycles as diagnosed by ultrasound and serum estradiol levels", Fertility and Sterility 37, 342, 1982.