| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Ovary |
a Epidemiology Branch
b Biostatistics Branch, National Institute of Environmental Health Sciences, Research Triangle Park, North Carolina 27709
c Westat, Research Triangle Park, North Carolina 27709
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
corpus luteum, corpus luteum function, human chorionic gonadotropin, pregnancy, progesterone
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The mechanisms of CL rescue vary among species [1]. In rodents, coitus initiates rescue. Even in the absence of fertilization, vaginal stimulation initiates hormonal changes that prolong the life of the CL. In primates, rescue is initiated by the conceptus through its production of chorionic gonadotropin (CG). In rhesus monkeys, for example, rescue is indicated by an abrupt 2- to 3-fold rise in serum progesterone that occurs within 24 h of implantation (defined by CG detection in maternal serum) [4, 5]. Progesterone remains elevated for several days and then declines rapidly to below midluteal levels. For the remainder of the pregnancy, the placenta is the primary source of serum progesterone. The luteotropic effects of CG have been demonstrated experimentally in nonpregnant primates, including humans [6–13]. Treatment of nonpregnant women with hCG on midluteal days of the menstrual cycle, when implantation would normally occur, prolongs the life of the CL and can also increase progesterone secretion [6, 7, 10–13].
The general pattern of progesterone production during the first trimester of human pregnancy has been described. Concentrations increase very early in gestation, decline in the fourth or fifth week after implantation, and then increase as the placenta becomes the primary source of progesterone [14, 15]. However, there are few data on change in progesterone levels around the time of implantation, data that would reflect the time of CL rescue [16–18]. The largest sample of pregnancies for which such data were recorded was presented by Lenton and Woodward [18]. Among the 18 conceptions they described, some pregnancies exhibited marked elevation in progesterone after implantation whereas others did not. However, their sample was too small for detailed analysis and included conceptions from infertility patients, which may not be representative of normal pregnancy.
The purpose of this study was to evaluate CL rescue and its role in early pregnancy loss in 120 naturally conceived human pregnancies, 43 of which were lost within 6 wk of the last menstrual period. We examined changes in daily urinary progesterone metabolite levels in the first week after implantation and assessed the relationship between early hCG levels and these progesterone metabolite profiles. The participants in the study had no known fertility problems. Women collected daily first-morning urine specimens while trying to conceive and during very early pregnancy. These specimens were used to measure hCG and the major urinary metabolite of progesterone. These measurements allowed us to identify the time of implantation and to characterize the progesterone metabolite profile during the first week after implantation.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The data were collected during a prospective study of 221 women attempting pregnancy, the North Carolina Early Pregnancy Study [19]. Women with no major chronic disease or known fertility problems collected and froze daily first-morning urine specimens beginning at the time when they discontinued their method of contraception in an attempt to become pregnant. Women provided informed consent, and the study was approved by the Institutional Review Board at the National Institute of Environmental Health Sciences.
One hundred ninety-nine pregnancies were conceived during the study: 151 clinical pregnancies (pregnancies that lasted for at least 6 wk past the last menstrual period) and 48 early losses (pregnancies lost within 6 wk of the last menstrual period). All 48 early losses and 101 clinical pregnancies were selected for analysis. The selected clinical pregnancies included the 15 pregnancies lost between 6 and 24 wk gestation and 86 live births, i.e., 12 pregnancies in the first 30 women enrolled, 11 pregnancies with 
1 day of spotting or bleeding during early pregnancy, 18 pregnancies in women who had a previous early loss in the study, and 45 additional pregnancies selected for completeness of urine collection. The hCG and progesterone metabolite levels were measured from daily specimens through the first week after implantation for clinical pregnancies and for the duration of elevated hCG for early losses.
The current analysis required that conception cycles have identifiable days of ovulation and implantation, preimplantation progesterone data for Luteal Days 5, 6, and 7 (so a reference level could be determined for each pregnancy), and no more than three progesterone values missing in the first postimplantation week. Seventy-seven of the 101 clinical pregnancies met these criteria (10 that spontaneously failed within 24 wk of the last menstrual period and 67 live births), as did 43 of the 48 early losses. A sample of 239 nonconception cycles from the study were available for comparison.
Hormone Assays and Hormonal Identification of Ovulation and Implantation
Daily collection of urine and urinary hormone assays allow for monitoring of hormonal changes over time in normal women; daily collection of blood was not feasible. Urine specimens were assayed for the major urinary metabolite of progesterone, pregnanediol-3-glucuronide (PdG), and for hCG. PdG was measured in duplicate or triplicate by RIA [20]. Human CG was measured in duplicate with an ultrasensitive polyclonal immunoradiometric assay designed to measure intact hCG (
and ß subunits in the dimeric form) [19, 21]. The antibody cross-reacted with the free ß subunit of hCG but not with LH or its subunits [22]. In other studies, urinary measures for hCG and PdG have been highly correlated with those of their counterparts in serum [23–25]. Urinary estrogen was measured in duplicate or triplicate by RIA [26] to aid in identifying day of ovulation.
Implantation was identified by the initial increase of urinary hCG. With a detection threshold of 0.01 ng/ml, the assay was sensitive enough to confirm the initial rise with pregnancy [27]. A day of implantation was assigned as the first day that hCG was sustained at concentrations exceeding 0.015 ng/ml [27].
Day of ovulation, Luteal Day 0 was identified by the rapid decline in the ratio of estrogen:progesterone metabolites. This decline reflects the abrupt fall in estrogen and rise in progesterone marking luteinization of the follicle, which occurs with ovulation [28]. This method is highly concordant with the use of midcycle surge of urinary LH and has comparable accuracy [28, 29]. Its precision is similar to that of serum LH determination [30]. The method has recently been validated in a large sample of cycles using ultrasound evidence of follicular rupture [31].
Quantification of Postimplantation Progesterone Metabolite Profiles
We systematically identified different patterns of postimplantation progesterone metabolite profiles by applying algorithms to detect rises and falls in PdG for individual women. Because concentrations of PdG in urine are dependent on metabolism, which varies among women, we examined changes relative to the preimplantation levels. The PdG concentrations on Luteal Days 5, 6, and 7 were averaged for the measure of preimplantation PdG. We identified a rise in PdG by looking for a sequence of PdG values that markedly exceeded the preimplantation level. The median coefficient of variation for preimplantation PdG levels was 15.5% in the 77 clinical conception cycles. We used this value as the standard for normal variability. The day of PdG rise was defined as the first day of a 2-day sequence in which values on both days were at least 31% higher (twice the median coefficient of variation) than the preimplantation value in that cycle. The algorithm was applied to 2-day sequences starting on the day of implantation.
Detection of a rise day was relatively insensitive to the 31% criterion. Lowering the criterion to 26% resulted in no additional rise days being detected. Raising the criterion to 36% resulted in one less rise day being detected. To test the extent to which this algorithm might identify random midluteal changes in PdG unrelated to pregnancy, we applied the algorithm to 239 nonconception cycles starting at Luteal Day 9 (Day 9 was the mean and median day of implantation for the 77 pregnancies, with a range of Day 8 to Day 12). A PdG rise was identified in 8% of the nonconception cycles (estimated false positive rate for the algorithm).
We also applied an algorithm to identify when progesterone had significantly declined. A day of waning PdG was defined as the second day of a 3-day sequence (starting no earlier than Luteal Day 8) in which values on all 3 days were >31% lower than the preimplantation PdG in that cycle. In the few cases of a single missing PdG value on a critical day, the mean of the days on either side was used as an imputed value. To assess the specificity of the criterion for waning PdG, we applied the algorithm to the 77 clinical pregnancies. Only one of the clinical pregnancies met the criterion in the first postimplantation week. The PdG in this pregnancy declined as in nonconception cycles, just meeting the criterion for waning PdG. Then the PdG recovered to midluteal levels, but it did not exhibit a PdG rise.
Several conceptions failed to exhibit either a PdG rise or a decline, so we included a category of PdG profile characterized by maintenance of preimplantation PdG levels during the first postimplantation week. The single clinical pregnancy that showed waning progesterone that then recovered to preimplantation levels was included in this category.
Two other measures of PdG response were used to describe the PdG profiles. The magnitude of the rise was defined as the average of the PdG on the PdG-rise day and on the following day divided by the preimplantation PdG. The PdG at the end of the first postimplantation week was defined as the average of the PdG on Days 6 and 7 after implantation divided by the preimplantation PdG.
Quantification of the hCG Signal
We used two quantitative markers of strength of the hCG signal: 1) the concentration of hCG on the day of implantation (analyzed as the natural log of the concentration because the distribution was highly skewed) and 2) the rate of early rise (defined as the slope of the log concentrations of hCG over the 3 days starting on the day of implantation).
Pregnancies ending in early loss showed a rise and then a fall of hCG. To describe the timing of conceptus failure in these pregnancies, we identified a day of hCG falter. Experimental studies indicate that maintenance of pregnancy does not depend on the absolute level of hCG but on a continuing increase. Therefore, falter was identified by a time when the hCG failed to increase. The algorithm for the day of hCG falter identified the first 4-day sequence in which the average percent change in hCG from day to day was 
0. In addition, none of the day-to-day changes within the 4-day sequence could be greater than or equal to a doubling of the concentration. The algorithm was applied starting on the day of implantation, and the falter day was defined as the third day of the 4-day sequence. In the few cases of a single missing hCG value on a critical day, the mean of the days on either side was used as an imputed value.
Statistical Procedures
The CL responses, as represented by the PdG profiles, were grouped into categories on the basis of presence or absence of a PdG rise or fall and the timing of a rise if one were present. The percentage of pregnancies in each category of PdG profile and rates of early pregnancy loss for each category were estimated for an unselected sample of pregnancies by weighting the conception cycles in our sample according to our sampling fraction of live births, clinical pregnancy losses, and early losses from the North Carolina Early Pregnancy Study.
Logistic regression was used to identify factors associated with PdG response. In modeling, the PdG profile categories were treated as polytomous outcomes. Conception cycles with early and late PdG rises were also compared separately. Significant relationships between the independent variables and the outcomes were based on the likelihood ratio chi-square. All P values given are two-sided.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Successful CL rescue occurred among the 77 clinical pregnancies. These conception cycles exhibited significantly higher progesterone metabolite levels than did nonconception cycles by the 10th day after ovulation (Luteal Day 10) (P < 0.05) (Fig. 1). When PdG data are analyzed by day of implantation, the average PdG concentration increased significantly on the day after implantation (P < 0.001) (Fig. 2) and continued to increase gradually during the first week after implantation (Fig. 2).
|
|
The gradual increase in mean PdG concentration after implantation suggests that humans do not exhibit the abrupt rise in progesterone described for nonhuman primates. However, an examination of PdG profiles in individual pregnancies reveals a different picture. Most clinical pregnancies showed an abrupt rise, but the rise occurred at various times after implantation. In other pregnancies, the PdG remained at preimplantation levels, exhibiting no increase throughout the 8-day window of observation starting at implantation (Fig. 3). Using the algorithm to detect a PdG rise, 60 of the 77 clinical pregnancies had an identifiable day of PdG rise. In the remainder of the pregnancies, PdG showed no marked increase above preimplantation levels.
|
To describe the variation in postimplantation PdG profiles, we grouped the clinical conception cycles into categories based on day of PdG rise: early rise (the PdG rise occurred within 2 days of implantation), late rise (the PdG rise occurred 3–6 days after implantation), and no rise (no PdG rise occurred during the first week after implantation). The mean daily change in PdG from preimplantation levels is shown for these three categories of PdG profile (Fig. 4). The early and late-rise categories both showed abrupt single-day increases of similar magnitude (average increase of 70% above preimplantation levels). PdG in the no-rise category was maintained near preimplantation levels throughout the first week. Thus, even the no-rise category had higher PdG levels than those in nonconception cycles (Fig. 4, bottom).
|
The PdG profiles during the first week after implantation for the 43 pregnancies that ended in early loss also showed heterogeneity. Seven exhibited early or late rises, 9 showed maintenance of PdG levels (no rise), and the remaining 27 were categorized as a PdG decline pattern (characterized by identifying a day of waning PdG within the first postimplantation week).
Table 1 shows the number of live births and pregnancy losses from our sample in each PdG profile category. When these data are weighted to take into account the analysis sampling frame, an estimated 45% of implanted pregnancies exhibit the early rise response, and the remaining pregnancies are distributed across the other three response categories (18% late rise, 22% no rise, and 15% decline). The few pregnancies that ended in clinical loss (miscarriage during Gestational Weeks 6–24), as for the other clinical pregnancies, were likely to have an early rise response, but the majority of early losses were in the decline category (Table 1). When the data are weighted by the sampling fraction, the estimated risk of early loss increased monotonically across the PdG profile categories ranging from 5% in the early rise category to 100% in the decline category (Fig. 5).
|
|
The PdG profile for early losses in a given category is similar to the PdG profile for clinical pregnancies in that category. The exception is that the early losses exhibiting early and late rise appear to have a less sustained PdG rise, i.e., the mean PdG levels at the end of the first postimplantation week are lower for early losses than for clinical pregnancies, although they still remain above midluteal levels.
Determinants of Postimplantation PdG Profile
Table 2 shows average preimplantation estrogen and progesterone metabolite levels and average early urinary hCG levels for the four PdG profile categories. Peak midcycle urinary estrogen and preimplantation PdG did not differ across the PdG profile categories, but urinary hCG levels did. Conceptions in the PdG decline category exhibited both lower initial urinary hCG concentrations and slower rates of early increase compared with the pregnancies in the other three categories. No differences among the other three groups were seen in rate of early urinary hCG increase, but pregnancies in the early rise category tended to have higher initial urinary hCG values than pregnancies in the late rise group.
|
Table 3 shows the timing of implantation for the four PdG profile categories. Late implantations were much more likely to be followed by declining PdG. However, few early implanting pregnancies were in the early rise category. Of the pregnancies implanting on Luteal Day 8, only 26% showed an early PdG rise, whereas 49% of pregnancies implanting on Days 9 or 10 had an early PdG rise.
|
Multivariate analyses confirmed the univariate relationships shown in Tables 2 and 3. Neither the midcycle urinary estrogen nor the preimplantation PdG differed by postimplantation PdG profile category. A low urinary hCG value on day of implantation, a slow early rise, and late implantation were independent predictors of PdG decline. The early-rise, late-rise, and no-rise categories did not differ significantly in rate of early hCG increase, but pregnancies in the late-rise category had a lower hCG on day of implantation than did pregnancies in the early-rise category (P = 0.05). Multivariate analyses also confirmed that Luteal Day 8 implantations had significantly fewer conceptions than expected in the early-rise category (P = 0.03).
Role of CL Failure in Early Pregnancy Loss
We examined the PdG profiles in relation to daily changes in urinary hCG in the early losses to look for further evidence of a role for CL failure in early loss. We first focused on the losses in the PdG decline category, where most of the losses occurred. Eighty percent of clinical pregnancies in the North Carolina Early Pregnancy Study implanted on Luteal Days 8–10 [27], so we reasoned that losses that implanted in that window would be the most viable conceptuses, and these would be the losses most likely to have factors other than a nonviable conceptus causing the loss. Figure 6 shows the day of waning PdG for losses after implantation on different days. The median day of waning PdG for the five losses that occurred after implantation on Luteal Day 9 was 3 days earlier than the median for nonconception cycles (Fig. 6), suggesting that some of these cycles may have had dysfunctional CL.
|
When individual conception cycles were examined, the day of waning PdG occurred before the day of hCG falter in four of the five losses that occurred in women that implanted on Day 9. One of these cycles appears to present a clear example of CL failure leading to loss (Fig. 7A). The progesterone had already waned by Day 9, the day of implantation, and menses began 2 days later. Although the hCG rose rapidly until menses, it then declined. This woman did not conceive a clinical pregnancy during her 6-mo participation in the study, and she had short luteal phases in all seven of her menstrual cycles.
|
In the remaining three early losses, in which implantation occurred by Day 9 and PdG waned early, the hCG profiles appear to be suboptimal, suggesting that the conceptus may have contributed to loss. For example, in one of these women (Fig. 7B), the early hCG rise was very slow, and the PdG decline occurred only a day earlier than average for nonconception cycles. However, after a 4- to 5-day delay, the hCG rose rapidly and continued to rise through menses before it too declined. The lack of a rapid early hCG rise may have contributed to this loss.
The seven early losses after implantation on Luteal Day 10 showed a slight shift to later days of waning PdG than expected, based on nonconception cycles (Fig. 6), and only two conception cycles showed decline of PdG before hCG falter. In both of these, the hCG rise appeared suboptimal.
We also examined losses in the no-rise category for evidence of inadequate CL rescue. Four of the nine conceptions in this group showed waning PdG before hCG falter. The clearest case of CL failure among these four is shown in Figure 7C. The conceptus implanted on Luteal Day 10, and the CL appeared to respond to the conceptus with prolonged production of progesterone. PdG was maintained at midluteal levels for 5 days after implantation but then declined despite continual increase in the hCG. The other three conceptions showed some disruption in hCG increase (example in Fig. 7D), suggesting that CL failure may not have been the only cause of early pregnancy loss. For the remaining 33 early losses, the majority of conceptuses implanted after the optimal implantation interval or hCG faltered before or nearly simultaneously with the day of waning PdG (examples in Fig. 7, E and F).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The variation in progesterone response to implantation may reflect the quality of the CL, uterine factors, or variation in the strength of the signal coming from the conceptus. To explore these possibilities, we examined markers of CL quality prior to implantation and characteristics of the daily urinary hCG profiles. We hypothesized that pregnancies with delayed or absent PdG rises show evidence of lower CL quality or weaker hCG signaling. Neither of the two markers of CL quality (peak estrogen metabolite levels, reflecting follicle development, and preimplantation PdG levels, reflecting luteinization) was related to the PdG profile. However, early hCG levels were important predictors. We also examined in detail the early losses to see when progesterone declined in relation to the timing of hCG falter. Evidence for CL failure was rarely found. Thus, our data suggest that the variability in progesterone response is more strongly associated with signals from the conceptus than with factors inherent to the CL. However, because hCG concentrations reflect both what the conceptus produces and what reaches the maternal circulation, a weak hCG signal could be caused by uterine problems as well as problems with the conceptus.
Lenton and Woodward [18] also examined determinants of CL response, although only 18 pregnancies were included in the analyses. They also reported a relationship between the strength of the hCG signal and progesterone response. Our findings do not support those of Lenton and Woodward that cycles with low preimplantation production of progesterone have a weaker progesterone response to pregnancy.
We also hypothesized that pregnancies implanting late would be associated with a weaker progesterone response because experimental data have shown a weak CL response when hCG is administered late in the luteal phase. Our data did show that late implantations were strongly associated with PdG decline but not with late rise or no rise. The pregnancies implanting early (Luteal Day 8) were less likely to have an early rise than those implanting on Luteal Days 9 or 10, a pattern also observed by Lenton and Woodward [18]. These authors interpreted this finding as evidence that the CL is unresponsive to hCG until later in the luteal phase. However, subsequent studies have shown that progesterone production, LH/hCG receptor expression, and response to exogenous hCG reach their peak during midluteal days, which include Day 8 [6, 11–13, 36, 37]. Thus, it is unlikely that the CL is unresponsive. Perhaps the conceptus, although producing hCG, is not mature enough to produce other signals important for stimulating a progesterone rise.
Other data also support the notion that hCG may not be the only important signal maintaining CL function during very early pregnancy. Other factors known to stimulate progesterone production by the CL in experimental studies include prostaglandins E2, D2, and I2 [38]. These prostaglandins might be produced during trophoblast invasion of the uterus. Molecular variants of hCG also may have distinct signaling functions [39–42], and the separate subunit of the hCG molecule may have some functional importance [43]. The assay for hCG used in our study was based on a polyclonal antibody developed to differentiate hCG from LH, but there was some cross-reactivity with the ß subunit of the hCG molecule. The antibody is no longer available, so its cross-reactivity to newly identifiable forms of hCG (such as glycosylated hCG) is unknown.
The CL increases production of substances other than progesterone during very early pregnancy, including estrogen, inhibin, and relaxin [44]. However, these hormones do not all increase at the same time. In primate species such as humans that have a prolonged gestational interval dependent on CL support, intact hCG may be the first of several signals to the CL. Signals demonstrating continued normal embryonic/trophoblast development may be required to maintain CL viability and to stimulate optimal production of the various substances required to maintain pregnancy.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
2 Correspondence: Donna Day Baird, Epidemiology Branch, Mail Drop A3-05, 111 TW Alexander Dr., Bldg. 101, Rm. 308, South Campus, National Institute of Environmental Health Sciences, Research Triangle Park, NC 27709. FAX: 919 541 2511; baird{at}niehs.nih.gov
Received: 13 June 2002.
First decision: 7 July 2002.
Accepted: 22 August 2002.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
-glucuronide. Acta Endocrinol 1980 93:228-233-glucuronide. J Steroid Biochem 1979 11:1165-1171[CrossRef][Medline]This article has been cited by other articles:
|
|
 |
|
  M. R. El-Talatini, A. H. Taylor, and J. C. Konje Fluctuation in anandamide levels from ovulation to early pregnancy in in-vitro fertilization-embryo transfer women, and its hormonal regulation Hum. Reprod., April 10, 2009; (2009) dep065v1. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P.A. Nepomnaschy, C.R. Weinberg, A.J. Wilcox, and D.D. Baird Urinary hCG patterns during the week following implantation Hum. Reprod., February 1, 2008; 23(2): 271 - 277. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Tong, L. Rombauts, A. Mulder, B. Marjono, J. L. Onwude, and E. M. Wallace Increased Day 15-17 Serum Pro-{alpha}C Inhibin Levels Specific to Successful Pregnancy J. Clin. Endocrinol. Metab., September 1, 2004; 89(9): 4464 - 4468. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
