If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Department of Reproductive Medicine, Oslo University Hospital, Oslo, NorwayDepartment of Gynaecology, Erasmus Medical Centre Rotterdam, Rotterdam, The NetherlandsCurrent address: Department of Otorhinolaryngology, HagaHospital, The Hague, The Netherlands
Institute of Clinical Medicine, Faculty of Medicine, University of Oslo, Oslo, NorwayDepartment of Pathology, Oslo University Hospital, Norwegian Radium Hospital, Oslo, Norway
Department of Reproductive Medicine, Oslo University Hospital, Oslo, NorwayInstitute of Clinical Medicine, Faculty of Medicine, University of Oslo, Oslo, Norway
How does follicle distribution evolve in the human ovarian cortex between the ages of 20 and 35 years?
Design
Fragments of ovarian cortex from women undergoing unilateral oophorectomy for fertility preservation were obtained for quantitative histological assessment, including recording the two-dimensional coordinates of the follicles. Data were analysed using spatial statistical methods.
Results
A total of 53 ovarian cortex tissue samples, containing 1–803 follicles each, were obtained from 14 women aged 20–35 years. Primordial and transitory follicles lay in a clustered manner in the human ovarian cortex, with an average cluster radius of around 270 µm (95% confidence interval 154–377 µm; n = 49). Follicle density declined with age (P = 0.006, n = 13), and the distance from the nearest neighbouring follicle increased (P = 0.004, n = 13). Cluster radius decreased with age (P = 0.02, n = 13), but the degree of clustering tended to increase (P = 0.11, n = 13). In the majority of the samples, follicles at different stages lay in different clusters (P < 0.05, n = 13).
Conclusions
This study shows that primordial and transitory follicles lie in different clusters in the human ovarian cortex. Spatio-temporal computer simulation suggests that interfollicular signals may hinder follicle loss and may therefore drive clustered follicle distribution. In clinical practice, the woman's age should be taken into account when assessing follicle density, as follicle distribution is increasingly clustered with advancing age.
). After this age, an even more rapid decrease in number is described, and when the stock declines to under a threshold of 1000 follicles, a woman enters menopause (
Prevailing models for the dynamics of reproductive ageing suggest that the rate of reproductive decline is related to the size of the remaining follicle cohort, so that the loss is accelerated when fewer follicles are present (
). Surprisingly, in women who had lost half of their ovarian reserve due to unilateral oophorectomy, the rate of decline was lower than predicted by the Faddy model (
). As primordial follicles are in close contact with each other in murine ovaries, the presence of a growth-suppressing substance that is released by the primordial follicles has been proposed (
). Such a factor might affect the global distribution of follicles during reproductive ageing. Indeed, a clustered distribution of follicles has been reported in the ovaries of aged mice (
). Feng and colleagues and Hu and co-workers applied the CLARITY method to render the whole murine ovary transparent and to assess follicular growth dynamics (
). Skodras and colleagues developed computer simulations to model the three-dimensional follicle distribution and test the precision of tissue sampling techniques (
Density and distribution of primordial follicles in single pieces of cortex from 21 patients and in individual pieces of cortex from three entire human ovaries.
). Poirot and co-workers and Qu an colleagues reported that follicles in the human ovarian cortex are not distributed homogeneously, although they did not further characterize the distribution pattern (
Density and distribution of primordial follicles in single pieces of cortex from 21 patients and in individual pieces of cortex from three entire human ovaries.
Intrigued by the challenges and controversy regarding ovarian ageing and follicle dynamics in women, the current study aimed to assess the spatial distribution of resting and early growing follicles in the human ovary over the period 20–35 years of age, when optimal fertility declines (
). There is no consensus on the distribution pattern, if any, of follicles in the human ovarian cortex but it may be of importance in clinical practice regarding ovarian tissue cryopreservation and transplantation.
Material and methods
Experimental design
This was an observational study.
Ethical approval
Approvals to use clinical data and biological samples were obtained from the Data Protection Officer (No. 19/6526 of 13-MAR-2019) and the Department of Pathology, Oslo University Hospital (No. 2019-09 of 04-MAR-2020). This project was included within the framework of quality assurance prior to every ovarian tissue auto-transplantation, and therefore it was exempt from obtaining separate informed consent and/or additional review from the research ethics committee.
Tissue source
For this study, ovarian tissue was obtained from women (n = 14) who were undergoing unilateral oophorectomy for tissue cryopreservation. During surgery, one ovary was removed, collected in transport medium (Dulbecco's phosphate-buffered saline [DPBS], D8662; Sigma-Aldrich, Merck KGaA, Germany) and shipped on ice to the laboratory at the Department of Reproductive Medicine. The organ was dissected under sterile conditions. An incision was made through the mesovarium and the medullar tissue was carefully removed using a sharp scalpel or pair of fine scissors. Once cleared from the medulla, the cortex was cut into fragments measuring approximately 5 × 5 × 1 mm3. The number of fragments collected varied (range 3–60, mean 28). The majority of the fragments were cryopreserved according to the protocol (
), and a few representative fragments (range 1–8, mean four) were used for histopathological assessment.
Sample preparation
The fragments that were assigned for histopathological assessment were fixed in 4% formalin. The tissue was dehydrated using ethanol, embedded in paraffin while oriented parallel with the ovarian surface and then sectioned using a microtome. The tissue sections were fixed onto glass plate, deparaffinized and stained with haematoxylin and eosin. The formalin, ethanol, paraffin, glass plates and haematoxylin–eosin stain were sourced from Sigma-Aldrich, Merck KGaA, Germany.
For the quantitative assessment of follicle distribution, paraffin-embedded tissue samples were sectioned again at 4 µm thickness at two different levels, exactly 250 µm apart. It was believed that this distance would prevent counting the same resting follicle in both sections while still allowing the assessment of a comparatively large cortical area before depletion of the paraffin blocks. This is in accordance with the findings of Charleston and colleagues (
The samples were scanned using the NanoZoomer XR scanner (Hamamatsu Photonics, Japan) and imported into the NanoZoomer Digital Pathology software, version 2.6.13 (Hamamatsu Photonics). This program was used for viewing and annotating the tissue samples as follows. First, the cortex was outlined using the freehand region annotation tool. All discontinuous fragments of cortex were outlined separately; for example, the tissue sample of patient 9 at level 1 yielded five separate cortical fragments. If present, the medulla was also outlined.
The image was then thoroughly searched for follicles; these were manually annotated using the pin tool, which planted a pin in the centre of the follicle. The stage of the follicle was annotated according to its histological features. Follicles that were surrounded by flattened pre-granulosa cells were considered to be primordial follicles. Follicles surrounded by a single layer of mixed flattened and cuboidal granulosa cells were considered to be intermediate follicles. Follicles surrounded by a single layer of cuboidal granulosa cells were considered to be in the primary stage. A follicle was assigned as secondary stage if multiple layers of cuboidal granulosa cells and fluid collections were present.
All objects of uncertain assignment were marked and reviewed by a trained pathologist (M.R.). After this review, structures that remained subject to discussion were further assessed by a specialist in gynaecological pathology (G.H.G.), who made a final judgement about whether to assign a structure as being a follicle.
Data analysis and statistics
The outline of the cortex and coordinates of follicles were imported into R software (
). The centres of the ovarian follicles were considered as a point pattern within a frame defined by the outline of the cortex.
The density of the follicles and the distance from the nearest neighbour were calculated with appropriate functions. To avoid bias because of the number of samples per patient, the data were analysed so that average follicle density for each patient was calculated. The individual data points were weighted by the total cortical area to take the sample size into consideration, as a larger sample area may mean more precise and reliable data.
The spatially random distribution of follicles was tested using two methods: the quadrat test, which compares quadrat counts with a random Poisson distribution, and the Clark–Evans test of aggregation, which considers a clustered point pattern as the alternative hypothesis. In addition, the aggregation index R was calculated for each sample, where R < 1 indicated a clustered point pattern. Clusters are defined by Knox as ‘a geographically bounded group of occurrences of sufficient size and concentration to be unlikely to have occurred by chance’ (
). The current study defined clusters as a group of follicles that are positioned closely together by visual assessment, surrounded by an area of lower follicle density.
The radius of point clusters was estimated using Ripley's L function, and the nadir of the first derivate was defined as the cluster radius (
). Ripley's L function represents the probability of finding another follicle at distance r from a given follicle in a homogeneous random distribution.
Weighted linear regression analysis was used to test the association between point pattern statistics and clinical data.
Follicle distribution model
Computer simulation was used to model the spatial and temporal dynamics of the primordial and transitory follicle patterns during the period 20–36 years of age, corresponding to the range in the observed data. This was performed to test different hypotheses for follicle dynamics.
As a model for the ovarian cortex, a two-dimensional rectangular frame of 50 × 50 mm was considered. The initial placement of points was determined by the simple sequential inhibition algorithm, which provides a random point distribution while never placing points closer than 45 µm. It was reasoned that the mean diameter of primordial follicles (44.6 ± 12.4 µm, n = 2005) would be the minimum allowed distance between the centres of two follicle-sized objects in the two-dimensional space.
Simulations consisted of 200 consecutive steps, each step corresponding to a 1-month period. At the start of a step, a probability of loss was assigned to each available point. If the probability of loss exceeded a set threshold (see below), the point would be eliminated and therefore be unavailable during subsequent steps. A total of 100 simulations per scenario were created on a high-performance computer and the average spatial point pattern across simulations were analysed.
The simulated scenarios were as follows. In scenario 1, the probability of loss of a point was linearly increasing with time, and was calculated as a random number drawn from the uniform distribution multiplied by time. This scenario simulated an accelerating follicle loss model in which neighbouring follicles did not interfere with each other. In scenario 2, the probability of loss was linearly increasing with time and also linearly increasing with distance greater than 256 µm (standard deviation 16 µm) from the nearest neighbouring point. In this scenario, the probability of loss also depended on whether or not a point was in the close vicinity of another point. This mimics the situation where the proximity of another follicle protects against the loss of the follicle, for example if a survival factor or growth-inhibiting factor is released by the follicles. The parameters of the simulation (the slope of the probability curve for time-related follicle loss and the threshold of distance from the neighbour with increased probability of loss) were extensively optimized across a wide range of possible values using the optim function in R, in order to maximize the fit of the model, i.e. minimize the weighted sum of squared deviations from the observed age versus follicle density data.
Results
Fifty-six samples of ovarian cortex derived from 14 women (1–6 samples per woman) undergoing ovarian tissue cryopreservation were analysed. In one patient, none of the three tissue samples contained follicles and therefore this patient was excluded from analysis.
The mean age of the included patients (n = 13) at the time of tissue collection was 27.9 years (range 20–35 years). None of the patients had ovarian pathology as an indication for ovarian tissue cryopreservation or had received chemotherapy before tissue collection. None of the patients had received radiotherapy to the pelvic region. The 53 cortical fragments that were assessed contained between 1 and 803 follicles, with an average of 47 follicles per sample. The characteristics of the patients and the samples are given in Table 1. An overview of all tissue samples and their details can be found in Supplementary Figure 1.
Table 1Characteristics of patients and ovarian tissue samples
In these patients, four cortical fragments contained only one primordial or transitory follicle. Therefore these four samples were excluded from the analysis.
In these patients, four cortical fragments contained only one primordial or transitory follicle. Therefore these four samples were excluded from the analysis.
In these patients, four cortical fragments contained only one primordial or transitory follicle. Therefore these four samples were excluded from the analysis.
This patient was excluded from the analysis as the tissue did not contain follicles.
25
1
0
a In these patients, four cortical fragments contained only one primordial or transitory follicle. Therefore these four samples were excluded from the analysis.
b This patient was excluded from the analysis as the tissue did not contain follicles.
Spatial distribution of primordial and transitory follicles
By visually analysing the fixed ovarian tissue (n = 53), follicles from the cortex, and in some samples potential follicle clusters, could be identified (Figure 1a). Four samples contained only one follicle and were therefore excluded from further analysis, as no pattern can be defined from a single point. In total, schematic representations from 49 tissues were obtained, including the extracted coordinates of follicles and the traces outlining the cortex (Figure 1b).
Figure 1Clustered distribution of primordial and transitory follicles in the human ovarian cortex. (A) Sample 16: ovarian tissue from a 24-year-old woman, which has been stained with haematoxylin and eosin; the inset shows a potential cluster of follicles. The right-hand image is a 10 × zoom-in of the boxed area. (B) Graphical representation of the ovarian tissue sample shown in panel (A). Individual follicles are indicated by black crosses. Follicle density is represented by a colour code: red represents a high follicle density and white a low density. (C) Ripley's L function chart (upper graph), which represents the probability of finding another follicle at distance r from a given follicle in a homogeneous random distribution (dotted red line) and the probability for the ovarian tissue sample from panel (A) (black curve). The size of the cluster can be estimated by the derivate of L (lower graph), where the first nadir indicates the cluster size. (D) Log P-values for hypothesis tests of spatial distribution of the follicles for all tissue samples (n = 49). A clustered distribution as an alternative hypothesis was assessed using the Clark–Evans test, and a spatially random Poisson distribution as an alternative hypothesis was assessed by the quadrat test.
The point patterns defined by the follicles for the presence of complete spatial randomness were tested with the quadrat and Clark–Evans tests. Using the quadrat test, 31 out of the 49 samples (63%) presented a significantly non-random distribution of follicles (P < 0.05; Figure 1d), whereas the Clark–Evans test for non-randomness indicated that 39 (80%) samples had a clustered follicle distribution (P < 0.05; Figure 1d).
The cumulative density of points within distance r from a given point was estimated by Ripley's L function. In cases with a clustered point pattern, L follows a characteristic shape with initially increasing values indicating a congestion of points in the cluster followed by decreasing values indicating the edge of the cluster (Figure 1c). The radius of the cluster can therefore be obtained from the derivate of L (
). Using this method for all the samples, a mean cluster radius of 266 µm was estimated (95% confidence interval 154–377 µm, n = 49 samples).
Distribution of primordial and transitory follicles
In order to compare the spatial distribution of the primordial and transitory follicles, 20 tissue samples with at least five follicles in each category were examined in six patients (Supplementary Table 1, Figure 2a, b), and the quadrat counts derived from this were compared using Pearson's chi-squared test. The minimum of five follicles per stage was chosen for statistical robustness. Thirteen (65%) samples had demonstrated significantly different (P < 0.05) quadrat counts and difference of L(r) from the theoretical random distribution (Figure 2c), indicating that groups of primordial and transitory follicles did not overlap and occupied distinct areas in the cortex.
Figure 2Primordial and transitory follicles tend to occupy different clusters. (A) Sample 31: ovarian tissue from a 27-year-old woman. (B) Representations of the sample shown in (A), crosses indicating individual follicles. Follicle density is represented by a colour code: red represents a high follicle density and white and yellow indicate a low density. (C) Ripley's L function chart, which represents the probability of finding another follicle at distance r from a given follicle in a homogeneous random distribution (dotted red line) and the probability for the ovarian tissue sample from panel (A). The black line indicates a significantly increased distance between the primordial and transitory follicles (with the shaded area showing the 95% confidence for no difference), suggesting that these follicles occupy different clusters.
). The study aimed to analyse the evolution of follicle distribution according to age. Using weighted linear regression, a decreasing density of follicles between was observed between 20 and 36 years of age (P = 0.006, n = 13; Figure 3a-d). In addition, increased age was associated with a reduced cluster radius (P = 0.02, linear regression), an increased distance from the nearest neighbour (P = 0.004, linear regression) and a tendency towards a reduced Clark–Evans R score (P = 0.11), indicating that loss of follicles during reproductive ageing was associated with clustering (Figure 3d).
Figure 3Association between follicle distribution and age. (A–C) Ovarian tissue fragments and follicle distribution representations from a 21-year-old woman, sample 1 (A), a 28-year-old woman, sample 29 (B), and a 35-year-old woman, sample 52 (C). Tissue fragments were stained with haematoxylin and eosin (upper panels). The lower panels are graphical representation of the same samples. Individual follicles are indicated by black crosses, and follicle density is represented by a colour code: red represents a high follicle density and yellow a low density. (D) Scatterplots indicating the association of age with, respectively, follicle density (P = 0.006), cluster radius (P = 0.02), nearest neighbour distance (P = 0.004) and Clark–Evans aggregation index R (P = 0.11). The size of the circles indicates the total area (mm2) of a tissue sample.
The emergence of follicle clusters during reproductive ageing may be a result of intra-ovarian paracrine factors that regulate the loss of follicles over a close distance. The effect of these factors was modelled by simulating follicle dynamics in a 50 × 50 mm two-dimensional space. Initially, this area was populated with 3000 points, achieving a final point density of 1.2 × 10–5 µm–2. Notably, the maximal density of follicles observed in women of 20 years of age was approximately 10–5 µm 2. The starting parameters of the simulation were derived from observed data, including the average radius of primordial follicles in all patients (44.6 ± 12.4 µm, n = 2005) and the density of primordial follicles (1.2 × 10–5 µm–2) at 20 years of age, which was the youngest age included in the study. The probability of follicle loss was defined as depending on closeness to other follicles and increasing progressively with time. Optimal fit with observed follicle density data was achieved when the survival of follicles depended on closeness to other follicles within 256 µm (standard deviation 16 µm) distance. The simulations were continued over 200 steps representing time intervals between 20 and 36 years of age. One hundred realizations of the simulated data were created.
The simulations affirmed that clustering increases with increased age, as indicated by the shape of the L(r) functions and the decreasing aggregation index R (Figure 4a-c). The simulations estimated the average values of cluster radius, the distance from the nearest neighbour, and the Clark–Evans aggregation index R within a reasonable range of the observed data (Figure 4c). The dynamics of cluster radius and nearest neighbour distance differed from the observed data, indicating an inferior fit of the simple model (Figure 4c). Simulations (n = 100) that only accounted for time and ignored the effect of closeness to other follicles consistently failed to yield a clustered point pattern distribution (Clark Evans aggregation index R = 1; Supplementary Table 1).
Figure 4Modelling of the spatio-temporal dynamics of primordial follicles in a simulated rectangular 50 × 50 mm tissue fragment. (A) Initial random follicle distribution at 20 years of age, and representative simulated distributions at 24, 28, 32 and 36 years of age. Red circles illustrate the respective cluster sizes, and the Clark–Evans aggregation index R and cluster radius r are given under the plots. (B) The Ripley's L function chart, which represents the probability of finding another follicle at distance r from a given follicle. In a homogeneous random distribution, the result (black curve) is very close to the dotted red line, which indicates complete randomness. With increasing age, the black line divides further from the red dotted line, indicating that the distribution becomes less random at higher ages. (C) Association between age and follicle density, cluster radius, nearest neighbour distance and Clark–Evans aggregation index R. Black circles indicate observed data, and the size of the circles indicates the total area (mm2) of a tissue sample. The mean of the simulated data is indicated by continuous red line, and the shaded area indicates the 95% confidence interval.
Ovarian ageing and follicle dynamics in the ovarian cortex are of considerable clinical interest. Insight into the complex physiological processes that take place in the ovary throughout reproductive life may be useful to improve fertility preservation treatment. Although the distribution of follicles in the ovarian cortex has been the subject of several studies, there has been no consensus on the distribution pattern, if any, of follicles in the human ovarian cortex. In this study, the relationship between follicle distribution in the human ovarian cortex and age was assessed by analysing follicle location in ovarian tissue samples from women undergoing ovarian tissue cryopreservation. The results provide insight into follicle dynamics and follicle distribution during the reproductive period (20–35 years).
It was found that, in most samples, ovarian follicles were localized in defined clusters with a radius of approximately 270 µm, with primordial and transitory follicles often occupying distinct clusters. Advanced age was associated with reduced follicle density and greater distance from the nearest neighbouring follicle. Moreover, there was a positive relation between age and the degree of clustering, while the radius of follicle clusters decreased with increasing age.
Primordial and transitory follicles followed a non-random distribution in the ovarian cortex, in agreement with several previous reports (
Density and distribution of primordial follicles in single pieces of cortex from 21 patients and in individual pieces of cortex from three entire human ovaries.
). Furthermore, Schmidt and colleagues analysed three whole human ovaries and reported an uneven distribution of follicles in all three cases, with follicles tending to lie in clusters (
Density and distribution of primordial follicles in single pieces of cortex from 21 patients and in individual pieces of cortex from three entire human ovaries.
In the current study, two different statistical assessments were carried out to evaluate the distribution of follicles. The majority of samples demonstrated a non-random follicle distribution (31/49, 63%) and follicle clustering by Clark–Evans test (39/49, 80%). Clustering of primordial follicles may contribute to sustained cyclical ovarian function during the reproductive transition (30–40 years), when follicle density is rapidly declining (
). Clusters may allow improved paracrine signalling compared with a random follicle distribution, and may thus promote the survival of primordial follicles and coordinated follicle growth. Indeed, it was observed in 13 of 20 (65%) samples that follicles in the growth phase were spatially separated from the resting cohort. Coordinated activation of the follicles may amplify a spatially clustered distribution, as entire clusters of follicles may remain dormant while others enter growth, and, as a result, clusters become increasingly recognizable. Feng and co-workers described how follicle clusters predominantly consisted of follicles of the same stage in mice, and that coordinated entry into the growth phase took place in these clusters (
). Furthermore, Da Silva-Buttkus and colleagues found that the chance of growth in a given primordial follicle in a murine ovary was lower within a 10 µm vicinity of another primordial follicle, suggesting that primordial follicles release an inhibiting signal that prevents growth in neighbouring follicles (
Density and distribution of primordial follicles in single pieces of cortex from 21 patients and in individual pieces of cortex from three entire human ovaries.
). Moreover, alterations in follicle distribution during reproductive ageing are described. For example, Faire and colleagues observed a homogenous distribution of follicles in Neonatal mice (1 week of age) (
). At a more advanced age, an increased degree of clustering was observed. Feng and co-workers assessed follicle distribution in young and aged mice, and confirmed an increased degree of clustering in aged animals (
). The current findings tend to support this observation by a correlation between age and increased clustering, as assessed by the Clark–Evans index R (P = 0.11).
Follicle-to-follicle inhibition may promote the maintenance of follicle clusters during a general depletion of ovarian reserve. Hypothetically, paracrine inhibitory signals released by resting follicles may prevent other primordial follicles from entering the growth phase. Such inhibitory signals could easily be transmitted to closely situated follicles in the cluster, thereby prolonging cluster survival and ovarian function. Previous reports suggest that cytokines from the transforming growth factor-β (TGF-β) superfamily may play a role in maintaining the dormant pool of follicles (
To further assess whether follicle-to-follicle signalling could contribute to a spatially clustered distribution in human ovaries, a computer simulation of ovarian follicle loss was created. In every successive step of the simulation, a probability of survival was assigned to each remaining follicle based on time and the distance from the nearest follicle. The characteristics and dynamics of the simulated tissue were similar to the observed data, indicating that signals among primordial follicles, with an effective range of a few hundred micrometres, do affect the rate of loss and spatial distribution of follicles.
The uneven clustered distribution of the follicles in the ovarian cortex may have multiple clinical consequences. First, as suggested in previous studies, the estimation of total follicle count may be increasingly imprecise and require more tissue samples when the distribution of follicles becomes clustered, which may already be notable at around 30 years of age. Second, the number of cortical strips that are reimplanted during ovarian tissue auto-transplantation should be adjusted to age (
Density and distribution of primordial follicles in single pieces of cortex from 21 patients and in individual pieces of cortex from three entire human ovaries.
). In the presence of clustering, there may an increased risk of grafting follicle-poor strips, which may require transplantation of multiple fragments. Third, the optimal size and thickness of tissue fragments may need to be considered, as the follicles are mostly clustered in an area of 500 µm. Fragments that are exceedingly thick may hinder the diffusion of cryoprotectants into the tissue (
) and result in follicle loss. Conversely, exceedingly thin fragments may disrupt clusters and prematurely deplete graft function, as seen by increased follicle loss in thin fragments of bovine ovarian tissue (
). Moreover, excessive dissection of ovarian tissue into small fragments may disrupt clusters that measure approximately 0.5 mm in diameter, which may lead to loss of inhibitory signals causing rapid successive waves of follicle activation and exhaustion of graft function. Therefore, careful assessment of the optimal biopsy and graft size should be carried out, making use of new knowledge regarding follicle distribution in order to optimize the chances of fertility preservation. Moreover, the existence of a paracrine factor acting between follicles that inhibits growth could be of interest in developing future techniques in the field of reproductive medicine. The targeting of such a factor could increase the number of growing oocytes and therefore the number of oocytes available for IVF treatment. Therefore, further research into the role of paracrine signalling of the TGF-β superfamily could be of clinical importance.
Several limitations of the current analysis can be recognized. The analysis involved small tissue fragments that were retained for diagnostic purposes, and the dimension, number and orientation of the samples were in line with clinical diagnostic routines. In particular, the small size of the available fragments hinders any validation of some basic assumptions of the spatial statistical analysis, including whether the samples were representative of the whole cortex with an identical distribution outside and inside the sample. Moreover, all the samples were sectioned in the same orientation, i.e. parallel to the ovarian surface. The distance between sections (250 µm) allowed overlapping of follicles and overestimation of follicle density to be avoided, but prevented a complete three-dimensional view of follicle distribution being obtained. Most previous studies have sectioned the tissue perpendicularly to the ovarian surface (which is the standard pathology practice) (
), but perpendicular sectioning would capture a small cross-section of the cortex and hinder the assessment of higher level distribution, including clustering. Moreover, there is a large variation in number of fragments analysed in different patients.
It is also acknowledged that the choice to exclude samples with only one follicle is arbitrary. One can argue that the minimal count should be set at a higher number. However, two follicles positioned very closely together, with no other follicles in the sample, could still be considered a ‘cluster’. Therefore, the choice was made to exclude all the samples that had only one follicle, but to include all other samples.
The computer model of follicle dynamics was based on a two-dimensional view of the ovary, which did not allow sufficiently tight packing of follicles in three dimensions to be achieved. Indeed, the simulation failed to estimate the observed nearest neighbour distance and cluster size, and it handled fewer follicles than would be possible using a three-dimensional model. Furthermore, the simplified simulation failed to account for more complex intra-ovarian effects, such as longer range stimulatory or inhibitory signals, as well as integrated effects of multiple neighbouring follicles. Collectively, this simple simulation may be ineffective to make predictions about trajectories of ovarian ageing beyond the observed data.
In conclusion, the distribution of primordial and transitory follicles in the ovarian cortex varies throughout the reproductive age. It was found that, in the majority of patients, the follicles lie in the ovarian cortex in a clustered manner. As the mean radius is approximately 270 µm, the mean diameter of the clusters at a young age is approximately 0.5 mm and shows a correlation with age, with a decreasing diameter with increasing age. In addition, there is a positive relationship between age and degree of clustering. Therefore, the results raise the suggestion of the presence of inhibitory factors acting among follicles within the same cluster and protecting the ovarian reserve. The presence of such a factor is conjectural based on the results of this study. Future research could focus on immunohistochemical staining experiments, to provide greater understanding of the intrafollicular environment.
As the majority of primordial follicles cluster together, the woman's age should always be considered when deciding on the number of samples that should be retrieved in order to record an accurate follicle count. Moreover, this study adds to the knowledge of follicle distribution in the human ovarian cortex, especially to the suggestion that these follicles lie in clusters. Finally, during auto-transplantation of ovarian cortex, the number, size and thickness of the tissue samples should be carefully adjusted, according to the age and follicle count.
Acknowledgements
The authors would like to thank Melinda Raki and Guro Horni Gløersen for their assistance with annotating the samples. They would also like to express their gratitude to the Department of Pathology for their help with scanning the samples.
Funding
A. Schenck was supported by the Erasmus+ Grant of the European Commission and the Gerrit Jan Mulder Travel Grant from the Erasmus Medical Centre Rotterdam during her stay in Oslo at the Oslo University Hospital. M. Vera-Rodriguez and P. Fedorcsák received funding from the Pink Ribbon Action/National Cancer Society.
Density and distribution of primordial follicles in single pieces of cortex from 21 patients and in individual pieces of cortex from three entire human ovaries.
Annejet Schenck visited the Department of Reproductive Medicine at Oslo University Hospital, Norway, for her research internship while she was a medical student at Erasmus University Rotterdam, the Netherlands. She has obtained her medical degree after completing that research and is currently a PhD student at HagaHospital, The Hague, the Netherlands.
Key Message
Based on 53 tissue samples from 14 women, it was concluded that follicles lie in clusters in the human ovarian cortex; in older age, clustering increases and clusters become smaller. This may indicate an inter-follicular signal hindering the loss of neighbouring follicles. The patient's age should be taken into account when assessing follicle density.
Article info
Publication history
Published online: October 25, 2020
Accepted:
October 22,
2020
Received in revised form:
October 7,
2020
Received:
July 31,
2020
Declaration: The authors report no financial or commercial conflicts of interest.
Footnotes
This work was presented as an e-poster at the virtual 36th Annual Meeting of ESHRE in July 2020.
30578-2&id=gr1.jpg){kind=link}
30578-2&id=gr2.jpg){kind=link}
30578-2&id=gr3.jpg){kind=link}
30578-2&id=gr4.jpg){kind=link}