^{1}

^{*}

^{2}

Conceived and designed the experiments: WHW TWK. Performed the experiments: TWK. Analyzed the data: WHW TWK. Wrote the paper: WHW TWK.

The authors have declared that no competing interests exist.

The human ovary contains a fixed number of non-growing follicles (NGFs) established before birth that decline with increasing age culminating in the menopause at 50–51 years. The objective of this study is to model the age-related population of NGFs in the human ovary from conception to menopause. Data were taken from eight separate quantitative histological studies (n = 325) in which NGF populations at known ages from seven weeks post conception to 51 years (median 32 years) were calculated. The data set was fitted to 20 peak function models, with the results ranked by obtained

Our current understanding of human ovarian reserve presumes that the ovary establishes several million non growing follicles (NGFs) at around five months of gestational age which is followed by a decline to the menopause when approximately 1,000 remain at an average age of 50–51 years

A number of recent reports have challenged this long held understanding of mammalian reproductive biology by reporting the presence of mitotically-active germ stem cells in juvenile and adult mouse ovaries

Several studies have reported the number of NGFs at different ages in humans

The highest ranked model (

The values for the parameters that maximise the

This model (illustrated graphically in

The best model for the establishment of the NGF population after conception, and the subsequent decline until age at menopause is described by an ADC model with parameters

The best model for the establishment of the NGF population after conception, and the subsequent decline until 25 years of age is described by an ADC model with parameters

To guard against model selection bias and to test the robustness of the model with respect to the data, we randomly removed 50 data points 61 times and re-fitted the models, with the ADC model being, on average, the best fitting model (double-sided t-test for difference of means,

To examine whether a model that permits neo-oogenesis would provide a better fit to the data, we further analysed the data by fitting models that need be neither asymmetric nor single peak. We found that any mathematical model that permits an increase in NGF population after the peak at 18–22 weeks has a markedly inferior fit compared to the best-fitting ADC model (

The highest-ranked non-peak model returned by TableCurve is a polynomial given by

If menopause is defined as a population of less than one thousand (in line with Faddy & Gosden

This figure gives illustrative examples of NGF populations predicted by our model. At ages 20 weeks, birth, 13 years, 25 years and 35 years the average NGF population is given, together with the respective 95% prediction intervals. The predicted average age at menopause (49.6 years) is also shown, together with the 95% prediction interval.

We describe the percentage of the NGF population remaining for a given age for women whose ovarian reserve is established and declines in line with our model (

The curve describes the percentage of ovarian reserve remaining at ages from birth to 55 years, based on the ADC model. 100% is taken to be the maximum ovarian reserve, occurring at 18–22 weeks post-conception. The percentages apply to all women whose ovarian reserve declines in line with our model (i.e. late and early menopause are associated with high and low peak NGF populations, respectively). We estimate that for 95% of women by the age of 30 years only 12% of their maximum pre-birth NGF population is present and by the age of 40 years only 3% remains.

This figure describes the hypothesis that individual age at menopause is determined by the peak NGF population established at around 20 weeks post-conception. The central curve is the ADC model described in

To investigate the number of NGFs recruited towards maturation and ovulation or apoptosis each month we have solved our model to show (

Each sub-figure describes the absolute number of NGFs recruited per month, for ages from birth to 55 years, based on population decline predicted by the ADC model.

In this study we have identified the first model of human ovarian reserve from conception to menopause that best fits the combined histological evidence. This model allows us to estimate the number of NGFs present in the ovary at any given age, suggests that 81% of the variance in NGF populations is due to age alone, and shows that the rate of NGF recruitment increases from birth to age 14 years then declines with age until menopause. Further analysis demonstrated that 95% of the NGF population variation is due to age alone for ages up to 25 years. The remaining 5% is due to factors other than age e.g. smoking, BMI, parity and stress. We can speculate that as chronological age increases, factors other than age become more important in determining the rate at which NGFs are lost through apoptosis.

We have made two major assumptions in our study. Firstly, that the results of the eight histological studies that have estimated the total number of NGFs per human ovary are comparable. The definition of a NGF is identical in six of the studies and similar in the remaining two studies. The counting techniques all used a variation of the technique first described by Block

Our second assumption is that the peak number of NGFs at 18–22 weeks gestation defines age at menopause for the individual woman, with early menopause women having low peak populations and late menopause women having high peak populations. The data on the number of NGFs in the ovary is cross-sectional: there is no longitudinal data available and in the absence of a non-invasive test to count NGFs in the individual woman this data is likely to remain unobtainable. Considered together the wide variation at age at menopause and wide variation of peak population of NGFs are suggestive but not conclusive evidence for this assumption to be tenable.

Since the publications by Johnson

We have described the percentage of the NGF population remaining for a given age for all women whose ovarian reserve is established and declines in line with our model (

Our finding that the rate of NGF recruitment increases to a plateau at just over 14 years and then decreases in all women irrespective of how many NGFs were established by birth is highly unlikely to be explained by coincidence. From the first comprehensive model of NGF decline from birth

In western society the average age of menarche is around 13 years

Can a more complete understanding of the establishment and decline of the non-renewing pool of NGFs help us to assess ovarian reserve for the individual woman? Several candidate markers for the assessment of ovarian reserve in the individual woman have been suggested including FSH, Inhibin B, AMH, and antral follicle counts and ovarian volume by transvaginal ultrasound

We have described and illustrated a model of human ovarian reserve from conception to menopause that best fits the combined histological evidence. Our model matches the log-adjusted NGF population to a five-parameter asymmetric double Gaussian cumulative (ADC) curve (

Data were taken from all eight known published quantitative histological studies of the human ovary (

Study | Statistics | |||||

1 | Bendsen | 2006 | 11 | −0.6 | −0.6 | −0.6 |

2 | Baker | 1963 | 11 | −0.6 | 7.0 | −0.2 |

3 | Forabosco | 2007 | 15 | −0.5 | 0.5 | −0.3 |

4 | Block | 1953 | 19 | −0.2 | 0.0 | 0.0 |

5 | Hansen | 2008 | 122 | 0.1 | 51.0 | 38.0 |

6 | Block | 1951 | 86 | 6.0 | 44.0 | 28.0 |

7 | Gougeon | 1987 | 52 | 25.0 | 46.0 | 39.5 |

8 | Richardson | 1987 | 9 | 45.0 | 51.0 | 46.0 |

Overall | 325 | −0.6 | 51.0 | 32.0 |

Study | Ovaries per range of ages | |||||||||||

1 | Bendsen | 11 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |

2 | Baker | 5 | 5 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |

3 | Forabosco | 14 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |

4 | Block | 5 | 14 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |

5 | Hansen | 0 | 6 | 3 | 4 | 8 | 8 | 8 | 12 | 22 | 26 | 15 |

6 | Block | 0 | 0 | 10 | 8 | 12 | 8 | 12 | 26 | 10 | 10 | 0 |

7 | Gougeon | 0 | 0 | 0 | 0 | 0 | 3 | 7 | 9 | 10 | 17 | 6 |

8 | Richardson | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 2 | 7 |

Overall | 35 | 26 | 14 | 12 | 20 | 19 | 27 | 47 | 42 | 55 | 28 |

We fitted 20 asymmetric peak models to the data set, using TableCurve-2D (Systat Software Inc., San Jose, California, USA), and ranked by

To avoid selection bias, we randomly removed 50 datapoints 61 times and re-fitted the models, calculating the mean and standard deviations of the

To calculate the rates of recruitment of NGFs towards maturation we solved the equations describing the early, late and average menopause models for all months from birth to menopause. The absolute numbers of NGFs recruited were then given by the differences in successive monthly totals.

This file contains (1) a plot of the ADC model that best fits the histological data, (2) full statistics for the model, (3) a plot of the residuals, (4) the dataset with values for the model, the residuals, the 95% CI for the model and the 95% prediction interval, (5) details of the iterations taken to obtain the model, and (6) the ranking of the peak functions as fitted to the dataset.

(0.18 MB PDF)

It may have been a chance result that the ADC performed well on the 325 datapoints selected, and that another model would perform better, in general, on similar datasets. The additional file contains details of the testing of this hypothesis, and shows that the ADC mean r^{2} was statistically significantly higher than any other peak model (p = 0:0065).

(0.05 MB PDF)

We would like to acknowledge the critical discussions we have had with Prof. Richard Anderson, Prof. Ian Gent, Dr. Jacob Howe, Dr. Dror Meirow and Dr. Evelyn Telfer.