Intracycle variation in number of antral follicles stratified by size and in endocrine markers of ovarian reserve in women with normal ovulatory menstrual cycles

Authors


Correspondence to: Dr S. Deb, Nottingham University Hospitals NHS Trust, Nottingham University Research and Treatment Unit in Reproduction (NURTURE), B Floor, East Block, Queen's Medical Centre, Derby Road, Nottingham, Nottinghamshire NG7 2UH, UK (e-mail: shilpa.deb@nuh.nhs.uk; shilpadeb@nottingham.ac.uk)

ABSTRACT

Objective

To quantify the intracycle variation in markers of ovarian reserve measured by antral follicle counts stratified by size using three-dimensional (3D) ultrasound and anti-Müllerian hormone (AMH) in women with normal menstrual cycles.

Methods

Healthy volunteers with normal menstrual cycles were prospectively recruited. Three-dimensional (3D) ultrasound examination and blood test were performed in early (F1) and mid-follicular (F2) phases and in periovulatory (PO) and luteal (LU) phases of one menstrual cycle. Antral follicles were measured using ‘sonography-based automated volume calculation’ with post processing (SonoAVC) and ovarian volume was measured using Virtual Organ Computer-aided AnaLysis (VOCAL). Blood serum was processed for hormonal assays including AMH, follicle stimulating hormone (FSH), luteinizing hormone (LH) and estradiol. Repeated-measures analysis was used to examine the variance in markers of ovarian reserve in different phases of one menstrual cycle.

Results

A total of 36 volunteers were included in the final analysis, of whom 34 attended all four visits. Repeated-measures analysis showed a significant variation in total antral follicle count (AFC) (P < 0.001). However, on stratifying the antral follicles according to size using SonoAVC, a non-significant variation (P = 0.382) was seen in small AFC (≤ 6.0 mm) and a significant variation (P < 0.001) was seen in large AFC (> 6.0 mm). The ovarian volume showed a significant intracycle variation (P < 0.001). A small but significant intracycle variation was noted in AMH (P = 0.041) and a significant variation was seen in levels of serum FSH, LH and estradiol (P < 0.05).

Conclusion

Small antral follicles (≤ 6.0 mm) measured using 3D ultrasound and AMH show little intracycle variation and perhaps should be evaluated when predicting ovarian reserve independent of menstrual cycle.

INTRODUCTION

The antral follicle count (AFC) is comparable to various multivariate models in the prediction of ovarian response to controlled ovarian stimulation during in-vitro fertilization treatment[1]. It involves counting antral follicles measuring 2–10 mm in both ovaries during the early follicular phase of the menstrual cycle defined as days 2 to 5[2-6]. Anti-Müllerian hormone (AMH), shown to be comparable to AFC in the prediction of ovarian response[7-12], also demonstrates a strong positive correlation with the number of small antral follicles[13]. Studies on reliability and validity of various tests of ovarian reserve are based on testing during the early follicular phase of the menstrual cycle. This relatively small window of opportunity is restrictive both to patients and to clinics performing the tests. An ideal test of ovarian reserve should be not only reliable but also independent of the menstrual cycle. Several groups have suggested stability in the levels of AMH throughout the menstrual cycle[14-17] whilst others have shown a presence of significant variation in the levels[18-20]. There are suggestions of a more profound intracycle variation in the number of antral follicles than in AMH levels[16]. As a result AMH is becoming the primary test of ovarian reserve.

Three-dimensional (3D) ultrasound assessment of the AFC has been shown to have high intra- and interobserver reliability[21-23]. Semiautomated follicle counts have recently become available and have also been shown to provide highly reliable measures of AFC[24] and their relative sizes[25]. Examining the variance in the markers of ovarian reserve will not only indicate the best time to perform these tests, but also help understand the expression of these markers in relation to the menstrual cycle. Sonography-based automated volume calculation (SonoAVC), shown as a reliable and valid method of assessing the size and number of antral follicles[24-26], was used to quantify the number of antral follicles stratified by size.

This study was designed to quantify intracycle variations in antral follicle counts of different size cohorts, ovarian volume, AMH, follicle stimulating hormone (FSH), luteinizing hormone (LH) and estradiol in normo-ovulatory healthy volunteers.

METHODS

Study participants described as ‘healthy volunteers’ were recruited prospectively through advertisements on the intranet portal of the University of Nottingham website and posters displayed in staff canteens, on notice boards and in the staff library. Further information about the study was communicated via email and information leaflets to those who expressed interest in participation. Inclusion criteria included age between 18 and 35 years, body mass index (BMI) between 18 and 26 kg/m2, regular menstrual cycles with a mean length ranging between 26 and 32 days, no history of ovarian surgery, no features suggestive of endocrine disease and no hormonal contraceptive use within the last 6 months.

Serial transvaginal ultrasound scans were performed and blood samples were taken during one menstrual cycle. With the first day of menstruation taken as day 1 of the cycle these visits were scheduled for the early and mid-follicular, periovulatory (days 12–16) and luteal (days 20–26) phases of the menstrual cycle. Periovulatory phase assessment was started on day 12 of the cycle, and the day of noting a dominant follicle of more than 16 mm was taken as the time point to include in the study. Luteal phase assessment was performed 7 days following the periovulatory study time point.

The study was approved by the ethics committee of the University of Nottingham, UK and conducted in accordance with ethical principles that have their origin in the Declaration of Helsinki on Ethical Principles for Medical Research Involving Human Subjects (1996), principles of good clinical practice and the Department of Health Research Governance Framework for Health and Social Care, 2005. Informed, written consent was obtained prior to the enrollment of each subject.

Ultrasound acquisition and analysis

Ultrasound scans were performed by a single investigator (S.D.) using a Voluson E8 Expert (GE Medical Systems, Zipf, Austria) and a three-dimensional 5–9-MHz, endovaginal transducer. Assessment involved a two-dimensional (2D) ultrasound scan of the pelvis to exclude any pelvic pathology. Our technique of volume ultrasound[24] included delineation of the ovary with application of a region of interest and the subsequent acquisition of a series of 2D planes acquired during a high quality, slow-sweep mode of the ultrasound beam through a predefined 90° angle. This ensured that the entire ovary and the greatest number of 2D planes were acquired, giving the highest degree of resolution when 2D data were reconstructed as a 3D volume.

Acquired 3D data were displayed in the multiplanar view of the ultrasound machine (Voluson E8 Expert). The gray-scale display of image was optimized and then rendered to generate a three-dimensional volume of interest (VOI). The render box was adjusted to exclude as much extraovarian information as possible and ensure that the entire ovary was included in the VOI. The threshold settings, which assign transparency associated with fluid to opaque voxels, were maintained for all datasets at a default setting of ‘low’. Once the dataset had been correctly positioned, 3D automated software SonoAVC was implemented[27]. The use of SonoAVC with postprocessing in counting and measuring the size of antral follicles has been previously described in detail[24]. In brief, the dimensions and relative sizes of individual follicles are displayed with a specific color. Postprocessing, involving the manual identification of follicles, was then used to ensure that all antral follicles were counted and measured. The mean ‘relaxed sphere diameter’, displayed as d(V), of each antral follicle in both ovaries was recorded and used for data analysis as this has been shown to most accurately reflect the true follicle diameter[28]. The antral follicle population for each subject was recorded starting from 2.0 mm up to a maximum of 10.0 mm.

Virtual Organ Computer-aided AnaLysis (VOCAL imaging program; GE Medical Systems) was used to quantify the volume of each ovary. The method, described in detail previously[29], involved manual delineation of the ovarian cortex in the B-plane of the multiplanar view as the dataset was rotated through a series of six consecutive 30° steps, and therefore a total of 180°, for each volume calculation.

Hormonal assays

Blood samples were centrifuged, within 30 min of collection, for 20 min at 4°C and 4000 rpm spin to separate the serum which was then frozen at –20°C and stored for subsequent analysis of AMH, FSH, LH and estradiol (E2) levels. A MIS/AMH enzyme-linked immunosorbent assay kit (Diagnostic Systems Laboratories, Webster, TX, USA) was used to measure serum AMH levels. The lowest detection limit was 0.006 ng/mL and the intra- and interassay coefficients of variation were below 5% and 8%, respectively.

The micro particle enzyme immunoassay was used to measure serum FSH, LH and estradiol levels on an AxSYM auto-analyzer (AxSYM; Abbott Laboratories, Abbott Park, IL, USA). The lowest detection limit for FSH was 0.37 IU/L, with intra- and interassay coefficients of variation below 5% and above 5%, respectively. The lowest detection limit for LH was 0.3 IU/L, with intra- and interassay coefficients of variation of 3% and 7%, respectively. The lowest detection limit for E2 was 8 pmol/L, with intra- and interassay coefficients of variation of 2.9–11% and 4.8–15.2%, respectively.

Statistical analysis

The Statistical Package for the Social Sciences (version 17.0; SPSS, Chicago, IL, USA) was used for statistical analysis. General linear model with repeated measures design was used to perform the analysis of variance. Test of sphericity was performed using Mauchly's test on the data. If the assumption of sphericity was violated, Greenhouse–Geisser correction was applied to the data and the P-value was derived subsequently. If the P-value derived following this correction was significant, the significance of the variation was reconfirmed using Hotelling's trace multivariate test and tests of within-subject contrasts. A P-value of less than 0.05 was considered statistically significant. Correlation coefficients were used to analyze the significance of correlation of the markers between the different cycle phases and intraclass correlation coefficients (ICC) were used to evaluate the true intraindividual intracycle variation.

RESULTS

A total of 38 healthy volunteers were recruited for the study. Two were excluded from the final analysis as a dominant follicle was not confirmed during ultrasound assessments. Of the 36 included in the final analysis, 34 attended all four visits. One subject did not attend the luteal phase visit and the other did not attend the second follicular phase visit.

Mean ± SD age and BMI of the participants was 28.12 ± 5.75 years and 22.34 ± 3.08 kg/m2, respectively. The mean ± SD of markers of ovarian reserve in different phases of the menstrual cycle are shown in Table 1.

Table 1. Comparison of ultrasound and endocrine determinants of ovarian reserve measured over one menstrual cycle in healthy volunteers with normal menstrual cycles. P-values were derived using repeated-measures analysis
ParameterPhase of menstrual cycleP
Early follicular (n = 36)Mid-follicular (n = 35)Periovulatory (n = 36)Luteal (n = 35)
  1. Data are given as mean ± SD. AFC, antral follicle count; AMH, anti-Müllerian hormone; FSH, follicle-stimulating hormone; LH, luteinizing hormone.

AFC for follicle size:
2.0–10.0 mm22.67 ± 9.9224.14 ± 10.0224.89 ± 9.8822.86 ± 9.83< 0.001
2.0–4.0 mm9.39 ± 4.969.17 ± 4.869.57 ± 4.989.81 ± 5.340.041
> 4.0–6.0 mm8.25 ± 3.158.42 ± 3.428.06 ± 3.368.19 ± 3.260.087
2.0–6.0 mm17.64 ± 7.2717.58 ± 7.4017.61 ± 7.5618.00 ± 7.680.380
> 6.0–10.0 mm5.03 ± 3.506.56 ± 3.467.28 ± 3.834.86 ± 3.57< 0.001
Ovarian volume (cm3)6.43 ± 2.196.30 ± 2.017.28 ± 2.467.67 ± 2.64< 0.001
AMH (ng/mL)2.61 ± 1.472.60 ± 1.392.61 ± 1.422.92 ± 1.660.041
FSH (IU/L)6.97 ± 2.436.54 ± 2.848.33 ± 2.914.97 ± 1.90< 0.001
LH (IU/L)5.84 ± 2.776.18 ± 2.7420.18 ± 17.055.50 ± 3.03< 0.001
Estradiol (pmol/L)147.72 ± 63.04165.25 ± 64.32893.25 ± 68.38427.78 ± 57.84< 0.001

Repeated-measures analysis of variance showed that the total number of antral follicles measuring 2.0–10.0 mm varied significantly (P = 0.002) across the menstrual cycle. Analysis of levels of contrasts showed a significant increase in count from the first follicular (F1) phase to the second follicular (F2) phase (F = 27.05; P < 0.001) and then a significant drop in count from the periovulatory phase (PO) to the luteal (LU) phase (F = 8.21; P = 0.007). There was no significant difference noted, however, in the total number of follicles measuring 2.0–10.0 mm between the F1 and LU phases of the menstrual cycle (P > 0.05) (Figure 1).

Figure 1.

Intracycle variation in mean antral follicle count (AFC), assessed using repeated-measures analysis, for follicles of size: (a) 2.0–10.0  mm (P < 0.001), (b) 2.0–6.0  mm (P = 0.380) and (c) > 6.0–10.0 mm (P < 0.001). Bars indicate standard error of the mean. Significant changes are indicated: **P < 0.01. F1, early follicular phase; F2, mid-follicular phase; PO, periovulatory phase; LU, luteal phase.

On stratifying the antral follicles into cohorts of small (2.0–4.0 mm and > 4.0–6.0 mm) and large (> 6.0–10.0 mm) antral follicles, we found no significant intracycle variation in the small antral follicles measuring up to 6.0 mm (P = 0.382) but did find a significant variation in larger follicles measuring more than 6.0 mm (P < 0.001). The test for levels of contrast with large antral follicles showed that the count significantly increases from the F2 phase to the PO phase (F = 36.32; P < 0.001) of the cycle and then drops significantly from the PO to the LU phase (F = 12.168; P = 0.001). The small antral follicles showed no significant variation at different levels of contrast (P > 0.05) (Figure 1).

Two subgroups of small antral follicles (2.0–4.0 mm and > 4.0–6.0 mm) were further analyzed. Whilst > 4–6.0 mm follicles showed no significant (P = 0.335) intracycle variation, the 2.0–4.0 mm follicles showed a small but significant (P = 0.042) intracycle variation, and this was noted between the first two follicular phases (F1 and F2) (F = 5.091; P = 0.033) (Figure 2).

Figure 2.

Intracycle variation in mean small antral follicle count (AFC), assessed using repeated-measures analysis, for follicles of size: (a) 2.0–4.0  mm (P = 0.335) and (b) > 4.0–6.0  mm (P = 0.042). Bars indicate standard error of the mean. Significant change is indicated: *P < 0.05. F1, early follicular phase; F2, mid-follicular phase; PO, periovulatory phase; LU, luteal phase.

The ovarian volume showed a significant intracycle variation (P < 0.001). This significance was noted because of the increase in ovarian volume between the F2 and PO phases (F = 9.44; P = 0.004) of the menstrual cycle, which was mainly attributable to the ovary containing the dominant follicle (Figure 3).

Figure 3.

Intracycle variation in mean ovarian volume assessed using repeated-measures analysis (P < 0.001). Bars indicate standard error of the mean. Significant change is indicated: **P < 0.01. F1, early follicular phase; F2, mid-follicular phase; PO, periovulatory phase; LU, luteal phase.

There was a small but significant intracycle variation noted in serum AMH levels (P = 0.041). On analysis of levels of within-subject contrasts, we found that AMH levels significantly increase between the PO and LU phases (F = 11.89; P = 0.039) of the menstrual cycle (Figure 4).

Figure 4.

Intracycle variation in mean: (a) anti-Müllerian hormone (AMH) (P = 0.041) levels, (b) follicle-stimulating hormone (FSH) (P < 0.001), (c) luteinizing hormone (LH) (P < 0.001) and (d) estradiol (P < 0.001), assessed using repeated-measures analysis. Bars indicate standard error of the mean. Significant change is indicated: *P < 0.05; **P < 0.01. F1, early follicular phase; F2, mid-follicular phase; PO, periovulatory phase; LU, luteal phase.

There were expected variations in serum FSH, LH and E2 levels. Serum FSH levels increased between the F2 and PO phases (F = 8.78; P = 0.005) of the menstrual cycle before falling between the PO and LU phases (F = 5 8.10; P < 0.001) when levels were significantly lower (P = 0.011) than during the F1 phase of the cycle. Serum LH levels showed a similar pattern, increasing significantly from the F2 to the PO phase (F = 23.98; P < 0.001) of the cycle before falling between the PO and the LU phase (F = 33.39; P < 0.001) when levels were comparable to those in the F1 phase of the cycle. Estradiol levels mirrored the gonadotropins, increasing from the F1 to the F2 phase (F = 115.48; P = 0.001) and from the F2 to the PO phase (F = 385.85; P < 0.001) before significantly decreasing between the PO and LU phases (F = 129.59; P < 0.001) where levels remained significantly higher than those during the F1 phase (Figure 4).

Whilst total AFC (2.0–10.0 mm), small (2.0–6.0  mm) and large (> 6.0–10.0 mm) antral follicle counts, ovarian volume and AMH showed significant correlation (r = 0.948, 0.976, 0.575, 0.568 and 0.959, respectively), estradiol, FSH and LH showed a non-significant correlation (r = 0.415, 0.408 and 0.373, respectively) between different phases of the menstrual cycle. AMH, total AFC (2.0–10.0 mm) and small AFC (2.0–6.0  mm) showed excellent ICC (0.96, 0.94 and 0.94, respectively), suggesting that the majority of intracycle variation is due to between-subject variation and that the true within-subject variation related to the phase of menstrual cycle was only 4%, 6% and 6%, respectively. The ICCs for larger antral follicles (> 6.0–10.0 mm), ovarian volume, FSH, LH and estradiol suggested high intraindividual intracycle variation (Table 2).

Table 2. Intraclass correlation coefficients (ICC) of ultrasound and endocrine markers of ovarian reserve over one menstrual cycle
ParameterICC95% CIP
  1. AFC, antral follicle count; AMH, anti-Müllerian hormone; FSH, follicle-stimulating hormone; LH, luteinizing hormone.

AFC for follicle size:
2.0–10.0 mm0.9390.901 to 0.965< 0.001
2.0–4.0 mm0.9310.890 to 0.961< 0.001
> 4.0–6.0 mm0.8980.839 to 0.941< 0.001
2.0–6.0 mm0.9380.900 to 0.965< 0.001
> 6.0–10.0 mm0.5180.353 to 0.681< 0.001
Ovarian volume (cm3)0.5880.431 to 0.734< 0.001
AMH (ng/mL)0.9570.931 to 0.976< 0.001
FSH (IU/L)0.3010.137 to 0.4930.012
LH (IU/L)0.104−0.031 to 0.2890.072
Estradiol (pmol/L)0.131−0.019 to 0.3200.058

DISCUSSION

This is the first study to examine antral follicles using the 3D ultrasound-assisted semi-automated technique SonoAVC. The results suggest a non-significant intracycle variation in number of small antral follicles (2.0–6.0  mm) but an excellent correlation between different phases of the menstrual cycle. The total AFC (2.0–10.0  mm) significantly varied across the menstrual cycle (P < 0.001). This was predominantly due to a variation in the number of larger antral follicles (> 6.0  mm) (P < 0.001) and, to a less but still significant extent, in the number of follicles measuring 2.0–4.0  mm (P = 0.041). There was no such variation in the number of follicles measuring 4.0–6.0  mm or in the number of small antral follicles measuring 2.0–6.0  mm as an overall cohort.

Our results suggest a small but significant intracycle variation in serum AMH (P = 0.041), mainly attributable to the increase in luteal phase levels. The increase in AMH levels in the luteal phase of the cycle with no concomitant change in the small antral follicle population seen in this study suggests that there might be a new recruitment of preantral and early antral follicles in the luteal phase[30-32] that cannot be identified on ultrasound. The increase in AMH we noted in the luteal phase has been described in one other study[14]. Hehenkamp et al. examined 44 healthy volunteers, with an average of seven visits per participant, and described the menstrual cycle from the mid-luteal phase of the previous cycle to the luteal phase of the next cycle[14]. The small significant increase in AMH noticed in the luteal phase might actually reflect the levels of the subsequent cycle, thereby raising the possibility that the number of early and preantral follicles expressing AMH may vary between cycles. The authors, however, believed that the luteal phase increase in AMH levels in their study could be non-significant due to the smaller number of measurements made and to a proportional increase in the number of younger patients assessed at that visit[14]. In our study we examined only four time points in the menstrual cycle and the small increase in luteal phase levels of AMH might have become non-significant with a higher number of visits during each cycle. This, however, may not have an impact on its use clinically as this small increase in the luteal phase may not inform treatment protocols in either the poor or the high ovarian reserve group . Wunder et al. showed a periovulatory increase in the levels of AMH in 36 women with normal menstrual cycles[20]. They, as we did in our study, described a period between two menstruations as one menstrual cycle. They examined the AMH level every other day and found that it increased in the late follicular phase and mid-luteal phase of the menstrual cycle, and that it decreased in the very early luteal phase, possibly because of the negative effect of luteinization on granulosa cells. The luteal phase increase in AMH levels in our study and the periovulatory increase in AMH levels shown by Wunder et al.[20] could be attributed to the preantral follicles not seen by ultrasound. It might be possible to explain the new recruitment of these follicles in the mid- to late luteal phase of the cycle when levels of gonadotropins are low, but it is difficult to explain the increase in AMH levels seen by Wunder et al. in the late follicular phase when the high levels of gonadotropins may have a negative impact on the expression of AMH. Cook et al. examined three time points (early follicular, periovulatory and luteal) in the menstrual cycle of 20 healthy women[18]. They found an increase in AMH levels in the periovulatory phase (LH surge + 1 day). These results are in contrast to what was found in our study and also to the results of Wunder et al.[20]. A small study population with fewer time points in the menstrual cycle may explain these findings.

Only one recent study has examined the intracycle variation in AMH along with the AFC made using 2D ultrasound[16]. Van Disseldorp et al. concluded that the variation in AMH during the menstrual cycle is significantly less than that seen in the AFC and that it was still a cycle-specific test of ovarian reserve. Moreover, they showed higher variation in the small antral follicles measuring 2–5  mm than in total AFC[16]. These results contradict our results, which suggest that the total AFC shows a significant intracycle variation and that the small antral follicles (2.0–6.0  mm) do not show significant intracycle variation. The difference in results may be due to the difference in methodology. In our study, antral follicles were counted and measured using SonoAVC which has been shown to be reliable in counting and measuring antral follicles. Also, 2D and other manual methods might overestimate the size of antral follicles[24, 25].

This is the first study that describes intracycle variation in ovarian volume using 3D ultrasound. It shows significant intracycle variation, especially that due to an increase in the periovulatory and luteal phases of cycle attributed to the dominant follicle and subsequent corpus luteum formation. Intercycle variation in ovarian volume has been described in subfertile women using 2D[33] and 3D ultrasound[34]. Ovarian volume must be assessed in the early follicular phase, i.e. before any significant follicle dominance occurs.

FSH and LH both showed significant intracycle variation, mainly due to the increase in the periovulatory phase and subsequent drop in the luteal phase of the menstrual cycle. This variation is well described[35] and therefore confirms that these tests are also best performed in the early follicular phase.

In conclusion, the ovarian reserve as measured using small AFC (2.0–6.0 mm) and AMH shows least intracycle variation and an excellent within-subject correlation. The small increase in the luteal phase levels of AMH might suggest a new recruitment of early antral and preantral follicles and, therefore, may more adequately predict the actual ovarian response but not necessarily the response to ovarian stimulation when compared to small antral follicles. Future studies designed to evaluate the ability to predict ovarian response following assisted reproduction treatment should possibly involve performance of these tests in different phases of the menstrual cycle.

Ancillary