- Email: [email protected]
Formation and early development of follicles in the polycystic ovary
ARTICLES
Formation and early development of follicles in the polycystic ovary
L J Webber, S Stubbs, J Stark, G H Trew, R Margara, K Hardy, S Franks
Summary Background Polycystic ovary syndrome is the most common cause of anovulatory infertility. It has long-term health implications and is an important risk factor for type 2 diabetes. However, little is known about the cause of polycystic ovaries. We have used detailed morphological analysis to assess the hypothesis that there is an intrinsic ovarian abnormality that affects the earliest stages of follicular development. Methods We took small cortical biopsies during routine laparoscopy from 24 women with normal ovaries and regular cycles and from 32 women with polycystic ovaries, 16 of whom had regular, ovulatory cycles and 16 of whom had oligomenorrhoea. We used computerised image analysis to assess the density and developmental stage of small preantral follicles in serial sections of fixed tissue. Findings Median density of small preantral follicles, including those at primordial and primary stages, was six-fold greater in biopsies from polycystic ovaries in anovulatory women than in normal ovaries (p=0·009). In both ovulatory and anovulatory women with polycystic ovaries, we noted a significant increase in the percentage of early growing (primary) follicles and a reciprocal decrease in the proportion of primordial follicles compared with normal ovaries. Interpretation Our findings indicate that there are fundamental differences between polycystic and normal ovaries in early follicular development, suggesting an intrinsic ovarian abnormality. The increased density of small preantral follicles in polycystic ovaries could result from increased population of the fetal ovary by germ cells, or from decreased rate of loss of oocytes during late gestation, childhood, and puberty. Lancet 2003; 362: 1017–21
Institute of Reproductive and Developmental Biology, Wolfson and Weston Research Centre for Family Health, Imperial College London, Hammersmith Hospital, London, UK (L J Webber MRCOG, S Stubbs MSc, G H Trew MRCOG, R Margara MD, K Hardy PhD, Prof S Franks FMedSci); and Department of Mathematics, Imperial College London, 180 Queens Gate, London (Prof J Stark PhD) Correspondence to: Prof S Franks, Wolfson and Weston Research Centre for Family Health, Institute of Reproductive and Developmental Biology, Imperial College London, Hammersmith Hospital, London W12 0NN, UK (e-mail: [email protected])
Introduction Polycystic ovary syndrome is the most common endocrine abnormality in women of reproductive age and is the main cause of anovulatory infertility. It is also associated with characteristic metabolic abnormalities, including insulin resistance, that are associated with an increased risk of type 2 diabetes1,2 and cardiovascular disease3 in later life. There are, therefore, important implications for women’s health. The cause of the syndrome remains uncertain; although, there is evidence that genetics have an important role.4 Clinical and biochemical features of polycystic ovary syndrome are typically heterogeneous,5–7 but the distinctive ovarian morphology seen in about a fifth of the female population, is, by most definitions, a characteristic feature of the syndrome.8–10 The major marker of the polycystic ovarian morphology is hyperandrogenaemia,11 and theca cells of polycystic ovaries have been shown to be the principal source of excess androgens.12,13 In anovulatory women with polycystic ovaries, the later (antral) stages of follicular development are clearly abnormal and growth of these follicles is typically arrested at a diameter of 5–8 mm—ie, well before a mature follicle would be expected to ovulate. This aberration of follicular maturation seems to indicate an abnormal endocrine environment,14 but whether there is an intrinsic abnormality of ovarian folliculogenesis remains unclear. Usually, growth of follicles from primordial to preovulatory stages takes 6 months or more15 and only the last 2 weeks of follicle development are highly dependent on cyclical changes in gonadotropins. Earlier in the process, preantral follicle development is primarily dependent on local growth factors and steroids of ovarian origin. These factors determine the survival and growth of a proportion of follicles but most are destined to die by atresia. In ovulatory women, only a few oocytes will eventually reach the preovulatory stage. Little is known about follicle formation and preantral growth in the polycystic ovary.16 We aimed to use detailed morphometric analysis of human ovarian cortical biopsies to test the hypothesis that there is an intrinsic abnormality of folliculogenesis in the polycystic ovary. We predicted that the early stages of follicle development would be abnormal in women with a history of anovulatory polycystic ovary syndrome and in those with polycystic ovaries but regular cycles.
Methods Participants Patients were recruited from the Hammersmith, Queen Charlotte’s and Chelsea, and St Mary’s Hospitals, London, UK, between December, 1997, and September, 2002. We obtained cortical ovarian biopsies from women who were having laparoscopic investigation or surgical treatment for infertility, or other surgical procedures. We established polycystic ovarian morphology or normal
THE LANCET • Vol 362 • September 27, 2003 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet.
1017
ARTICLES
ovarian morphology with ultrasonography. We did ultrasonography, measured mid-luteal progesterone concentrations, or both for evidence of ovulation. Serum testosterone and gonadotropins were measured by specific radioimmunoassay with methods described elsewhere.9 Serum testosterone was not routinely measured in ovulatory participants but we did measure concentrations of this hormone in anovulatory women with polycystic ovaries, We obtained serum in the early follicular phase (days 2–4 of the cycle) to measure gonadotropin concentrations. Cortical tissue was taken from the surface of the ovary opposite the hilum either with a biopsy device (Cook, Hertfordshire, UK) during laparoscopy or with a scalpel during a laparotomy. In women with regular cycles, the biopsy was, in most cases, taken in the mid-luteal phase. If a dominant follicle or corpus luteum was present, we took a sample from the contralateral ovary. Biopsies had a maximum volume of 5 mm3. The study was approved by the research ethics committees of the Imperial College School of Medicine in Hammersmith, Queen Charlotte’s and Chelsea and Acton Hospitals, and the St Mary’s NHS Trust. Patients gave written consent. Tissue preparation and histological assessment Ovarian tissue was gathered into HEPES-buffered minimum essential medium for immediate transfer at 37°C to the laboratory, where it was divided into pieces less than 1 mm in diameter under sterile conditions. One control piece was fixed in Bouin’s solution, stored in alcohol, and later dehydrated and blocked in paraffin wax. Serial sections, 5 m thick, were cut from the blocks, mounted on microscope slides, and stained with haematoxylin and eosin. We prepared about 24 slides, all containing a maximum of seven sections, for each patient. Samples were anonymised and coded so that the observer (LJW) was unable to identify the type of ovary from which the biopsy was taken until data collection was complete. Assessment of follicle development and survival Slides were examined either on a standard Olympus light microscope with a ⫻63 objective or on a Nikon Eclipse E600 light microscope (⫻60 objective) by one observer (LJW). Images were captured with a DXM1200 digital camera (Nikon, Kingston-upon-Thames, UK) with LUCIA image analysis software (Nikon). The number of follicles in each of the tissue pieces was counted and their stage of development and health were assessed. To ensure the most accurate assessment, we examined each follicle in every section on which it appeared. The stage of follicle development was assessed as follows: those with one layer of flattened pregranulosa cells were judged to be primordial; when there was a single layer of cells in which most were expanded cuboidal granulosa cells, the follicle was described as primary. Secondary follicles had more than one layer of granulosa cells, but if there were more than five layers, we classified the follicle as multilayered preantral.17 We judged follicles as healthy or atretic on the basis of morphology. A degenerated oocyte nucleus, uneven or folded nuclear membrane, vacuoles in the oocyte or pyknotic nuclei of several granulosa cells were all regarded as signs of atresia (degeneration) of follicles.18 Atretic follicles could not be staged accurately. Follicle diameter was measured with on-screen calipers provided by the LUCIA software. A second observer (SS) reanalysed a number of sections to confirm accuracy and reproducibility of measurements. Tissue volume was measured to allow the calculation of follicle density—ie, follicle count per mm3 of tissue. The
1018
area of every fifth section of tissue was digitally measured by image analysis (LUCIA image analysis software). We then calculated volume as mean section area⫻number of sections ⫻5 m (ie, section thickness). Statistical analysis We compared ages of women with polycystic ovaries with those of women with normal ovaries using ANOVA: significance was judged to be p⭐0·05. We compared follicle densities in ovarian tissue samples (⭓0·1 mm3) with the Kruskal-Wallis test (non-parametric ANOVA) using the GraphPad InStat version 3.0a for Macintosh (GraphPad Software, San Diego, California USA). If significant, paired comparisons were done with the Mann Whitney test and a Bonferroni correction. Proportions of resting (primordial) and growing follicles in each of the groups (normal, ovulatory polycystic ovaries, and anovulatory polycystic ovaries) were analysed with binomial regression, allowing for over-dispersion. Analysis of proportions is complicated by the fact that different samples had varying numbers of follicles. As a result, estimates of proportions in individual samples varied widely in their accuracy, with samples that had a large number of follicles providing more accurate estimates. For example, in a sample containing only two follicles, the proportion could only be one of 0%, 50%, or 100%. Unless the true value was one of these, the estimate could be substantially inaccurate. Furthermore, even if the true value were close to one of 0%, 50%, or 100% we would still expect to see large errors because of the random effects of sampling. Thus, estimates of proportion, and their comparison between different groups should take into account the number of follicles in each sample. The easiest way to take this variable into account is to assume that each follicle in a sample behaves independently. However, this assumption is unlikely to be true because we can expect correlations between the follicles taken from the same sample, and might result in a phenomenon called over-dispersion—ie, variability in the observed proportions is much greater than that predicted by a binomial distribution. There are several standard ways of addressing this issue. We used one of the most widely used methods, and estimated the dispersion of the observed data and then scaled the SE by this amount. Estimates of proportions, 95% CI and significance levels were all done with the binary regression command in Stata8 (Stata Corporation, College Station TX, USA). This tool calculates a maximum likelihood (or minimum deviance) estimate, computes SE with the resulting information matrix, and then scales them using a deviance-based estimate of the over-dispersion. Role of funding source The sponsors of this study had no role in study design, data collection, data analysis, data interpretation, or the writing of the report.
Results 40% of patients that we asked agreed to participate, and 56 women donated cortical ovarian samples. 32 had polycystic ovarian morphology, and 24 had normal ovaries. All women were having investigation or surgical treatment for infertility, apart from three who were undergoing reversal of sterilisation and one who was having a hysterectomy. All those with normal ovaries and 16 of those with polycystic ovaries had evidence of ovulation, as shown by ultrasonography, normal midluteal progestorone concentrations, or both. The remaining 16 women with polycystic ovaries had
THE LANCET • Vol 362 • September 27, 2003 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet.
ARTICLES
Number of follicles/mm3
100 80 60 40 20 0 O O al rm y PC y PC o N tor tor a a ul ul Ov nov A
O O al rm y PC y PC o N tor tor a a ul ul Ov nov A
O O al rm y PC y PC o N tor tor a a ul ul Ov nov A
Total Primordial Primary Figure 1: Median density of healthy and atretic preantral follicles in cortical biopsies PCO=polycystic ovaries. Biopsy samples were ⭓0·1 mm3. Samples taken from 22 women with normal ovaries, 14 ovulatory women with polycystic ovaries, and 15 anovulatory women with polycystic ovaries. Bars are 95% CI.
100 80 Follicles (%)
oligomenorrhoea or amenorrhoea, were known to be anovulatory, and had clinical or biochemical evidence of hyperandrogenism (ie, they fitted the classic definition of polycystic ovary syndrome10). Mean age did not differ between groups: normal: 32·9 years (SE 1·0); ovulatory polycystic ovaries: 32·4 years (1·1); anovulatory polycystic ovaries: 30·2 (1·4) (p=0·2, ANOVA). Ovarian volume varied significantly between the three groups (p=0·002, ANOVA). Mean volume, calculated from ultrasonograph measurements,9 was greater in anovulatory polycystic ovaries (11·0 mm3 [1·2]) than in either ovulatory polycystic ovaries (7·5 mm3 [0·8]) (p<0·05) or controls (6·0 mm3 [0·5]; p<0·01, posthoc Tukey-Kramer multiple comparisons test). Mean serum testosterone concentrations in anovulatory women with polycystic ovaries was higher than that in a reference group of 50 women with normal ovaries and regular cycles (2·5 nmol/L [0·6] vs 1·4 nmol/L [0·1]; p<0·0001, Student’s unpaired t test). Serum concentrations of luteinising hormone in anovulatory women with polycystic ovaries did not differ from those in ovulatory women with polycystic ovaries and controls (7·3 IU/L [1·2]; 5·7 IU/L [1·0]; and 5·6 IU/L [0·7], respectively). Serum folicle-stimulating hormone concentrations were similar in all three groups (5·6 IU/L [0·3]; 5·5 IU/L [0·6]; and 6·5 IU/L [0·5] IU/L). None of the participants had a raised concentration of serum folicle-stimulating hormone (>12·8 IU/L). More than 90% of follicles in biopsies from both polycystic and normal ovaries were at either the primordial or the primary stage. Secondary follicles were occasionally seen, but only one early antral follicle was noted (from a normal ovary) and there were no multilayered preantral follicles. There were, therefore, too few secondary and preantral follicles to allow accurate calculation of their follicle density. Median follicle diameter was 40·2 m (range 20–105). Median volumes of tissue analysed for the three groups were not significantly different for ovulatory polycystic ovaries, anovulatory polycystic ovaries, or normal ovaries (0·40 mm3 [0·05–0·92], 0·38 [0·09–4·18], and 0·44 mm3 [0·09–3·06], respectively). Follicle densities in tissue from polycystic and normal ovaries are shown in figure 1. Density of the total population of small preantral follicles differed between groups: normal ovaries 11·4 follicles per mm3 (4–34); ovulatory polycystic ovaries 27·4 follicles per mm3 (9–81);
60 40 20 0 al
O
rm
No
at
ul
Ov
y or
PC
at
ul
v no
al
O
PC
y or A
Primordial
O
rm
No
at
ul
Ov
O
PC
y or
PC
at
ul
v no
y or
A
Growing
Figure 2: Mean proportion of total healthy small preantral follicles at the primordial and primary stages PCO=polycystic ovaries. Bars are 95% CI.
anovulatory polycystic ovaries 73·0 follicles per mm3 (31–94) (p=0·009, Kruskal-Wallis). The density of primary follicles also differed significantly between groups (p=0·006, Kruskal-Wallis). When tissue from women with polycystic ovaries was classified by ovulation status, anovulatory polycystic ovaries had a higher overall density of follicles than did normal ovaries (p=0·009, Mann Whitney with Bonferroni correction) but there were no significant differences in density between anovulatory and ovulatory polycystic ovaries (p=0·38) or between ovulatory polycystic ovaries and normal tissue (p=0·4). Primordial follicle density did not differ between the three groups (p=0·3, KruskalWallis) but the difference in density of primary follicles was significant (p=0·006, Kruskal-Wallis) and, as with total follicle density, the major difference was between anovulatory polycystic and normal ovaries (p=0·004, Mann Whitney with Bonferroni correction). We did not note significant differences between the primary follicle densities in normal and ovulatory polycystic ovaries. The proportion of healthy primordial follicles was lower in both types of polycystic ovaries than in normal ovaries (anovulatory polycystic ovaries, p=0·001; ovulatory polycystic ovaries, p=0·03) (figure 2). Compared with normal ovaries, the proportion of healthy primary follicles that had started growing was significantly higher in anovulatory (p=0·001) and ovulatory (p=0·03) polycystic ovaries. We did not note a significant difference in proportions between the two polycystic ovary groups (p=0·7). In normal ovaries, the mean proportion of atretic follicles was 19% (95% CI 14–27), 28% (22–35) in polycystic ovaries from anovulatory women, and 23% (14–35) in ovulatory polycystic ovaries. Although the difference between anovulatory polycystic and normal ovaries approached significance (p=0·06; binomial regression), proportions did not differ between ovulatory polycystic and normal ovaries (p=0·59), or between the two polycystic ovary groups (p=0·38).
Discussion Median density of small preantral follicles, including those at primordial and primary stages, was six-fold greater in anovulatory polycystic ovaries than in normal ovaries. In ovulatory and anovulatory women with polycystic ovaries, the proportion of early growing (primary) follicles was higher than that in normal ovaries, and we noted a reciprocal decrease in the proportion of primordial follicles.
THE LANCET • Vol 362 • September 27, 2003 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet.
1019
ARTICLES
compared with normal ovaries. Indeed, the trend was towards a higher proportion of atretic follicles in anovulatory More PGC More cell More cells polycystic ovaries than in normal ovaries. entering divisions? enclosed by These data suggest that the dynamics of ovary? pregranulosa early follicular development in polycystic cells? ovaries are quite distinct from those in the Increased density normal ovary. Polycystic ovary of small preantral Increased recruitment of follicles from Normal ovary follicles in polycystic the resting pool to the early growing ovaries phase might be expected to encourage Slower rate premature depletion of the stock of of atresia in primordial follicles in polycystic ovaries polycystic and, therefore, to accelerate the onset of ovaries? menopause. However, there is little evidence that menopause in women with polycystic ovaries occurs earlier than in the general population.2,21 The explanation for this apparent absence of effect on age of menopause might be that the increased recruitment from the resting pool is counterbalanced by the higher initial number of follicles in that pool. An alternative (or additional) factor to be considered is that the rate of progression through the later stages of preantral follicle development could be slower in Primordial Mitosis of Follicle Atresia germ cells germ cells formation the polycystic ovary. In many respects, our data are Figure 3: Fate of germ cells in the human fetal ovary and proposed mechanisms to consistent with the findings of explain the increased density of small preantral follicles in the polycystic ovary Hughesdon16 who made a detailed After migration of primordial germ cells into the developing ovary, the population of germ cells histological comparison of sampled increases by mitosis to reach a maximum of seven million at around 20 weeks’ gestation. sections, comprising cortical, subcortical, Follicle formation (the surrounding of primary oocytes by a single layer of somatic, pregranulosa and medullary tissue, from archived cells) might begin as early as the 12th week in utero. The population of germ cells (both naked normal and polycystic ovaries. Although and in primordial follicles) declines steadily from midgestation, by the process of atresia, to reach a level of around two million by birth. our biopsies were confined to cortical tissue, the focus in our study was on early follicle development and the distribution of primordial Infertility resulting from anovulation is one of the most and primary follicles is restricted to the outer 1–2 mm of common features of polycystic ovary syndrome5–7 and the cortex.22 Hughesdon observed that there was some syndrome is the most frequent cause of anovulation. Our data suggest that abnormalities of folliculogenesis that subcortical dislocation of small preantral follicles (ie, lead to anovulation in the polycystic ovary arise at the primordial and primary) but this was more common in earliest stages of follicle development. There are four polycystic ovaries so that, if anything, our measurements possible explanations for the higher overall density of of the density of primordial and primary follicles in small preantral (and specifically of primary) follicles in ovarian cortex would underestimate differences between polycystic ovaries from anovulatory women (figure 3). groups. Hughesdon noted that the number of growing First, the fetal ovary might be endowed with an increased follicles at all stages (including preantral) was increased population of primordial germ cells. The second in polycystic ovaries. However, he reported that the possibility is that the oogonia undergo more mitotic number of primordial follicles per section was much the divisions in the fetal ovary. Third, follicle formation, same in normal and polycystic ovaries in his study. This which begins at about 12 weeks’ gestation20 and involves finding could reflect the fact that in our study we have specifically examined follicle density using serial, rather assembly of somatic cells around the naked oocytes, might than sampled, sections and, in addition, we were able to be enhanced in the polycystic ovary. Fourth, the high rate classify participants by ovulatory status. of loss of germ cells19 (and their surrounding somatic cells) Ample evidence exists to suggest that polycystic ovary that begins in the fetal ovary and continues until the time syndrome has a major genetic cause.4 As yet, the gene (or of the menopause,15 might be reduced in the polycystic ovary. more probably genes) that contribute to the polycystic The increased density of follicles in adult polycystic ovary phenotype remain uncertain. These observations ovaries suggests that, especially in the case of women with lend support to the view that genes involved in early chronic anovulation, such ovaries are characterised by a follicular growth are candidates in the cause of polycystic greater initial pool of follicles. Analysis of the proportions ovary syndrome. Little is known about the molecules that of primordial (resting) and primary (growing) follicles control recruitment and early growth of follicles, but indicates that, irrespective of ovulation status, there might growth factors in the transforming growth factor beta also be increased recruitment (initiation of growth) from (TGF) family such as anti-Mullerian hormone (AMH)23 the resting follicle pool in polycystic ovaries. The and growth differentiation factor-9 (GDF-9)24 have been proportion of atretic follicles in ovulatory or anovulatory implicated and so have androgens.6 Serum concentrations polycystic ovaries did not differ from that in normal of androgens are typically high in women with polycystic ovaries, indicating that the difference in follicle density is ovary syndrome. Treatment of adult Rhesus monkeys not attributable to a reduced rate of atresia in adulthood with androgens has been shown to increase recruitment Number of germ cells
? Higher initial population of primordial follicles
19
20
19
1020
THE LANCET • Vol 362 • September 27, 2003 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet.
ARTICLES
and stimulate growth of preantral follicles.25 The mechanism of androgen action remains unclear, but Vendola and colleagues25 have suggested that the intraovarian insulin-like growth factor (IGF) system is involved; androgens stimulate expression of IGF-1 and its receptor in early growing follicles. Importantly, prenatal exposure to androgens in Rhesus monkies and sheep produces endocrine changes that are very similar to those in polycystic ovary syndrome.26 Androgens might, therefore, play a part in abnormal early folliculogenesis which could, in turn, lead to abnormal function of theca cells in mature follicles, thus setting up a self-propagating cycle of abnormal follicular growth and function.26 Thus, our findings pave the way for further molecular and cellular studies into the origin of this hitherto puzzling syndrome. We have shown abnormalities of folliculogenesis in both ovulatory and anovulatory polycystic ovaries. The disorder is more pronounced in anovulatory polycystic ovaries and might, in part, contribute to the failure of the development of a dominant follicle and subsequent ovulation. Our findings provide more evidence for the central role of an ovarian abnormality in polycystic ovary syndrome. Contributors L Webber did laboratory studies (including histological analyses) as part of her PhD project and was primarily responsible for recruiting patients and obtaining written consent. S Stubbs assisted in histological analyses. G Trew and R Margara supervised recruitment of patients and took biopsy samples. J Stark did statistical analysis. K Hardy and S Franks conceived and supervised the project.
6 7
8
9 10
11
12
13
14
15 16
17
Conflict of interest statement None declared.
18
Acknowledgments This work was funded by Programme Grants from the Medical Research Council (salary for LJW and laboratory expenses) and Wellbeing (salary for SS and laboratory expenses). We thank Catherine S Wright who collected and prepared some of the tissue samples that were used in these analyses.
19 20 21
References 1
2
3
4 5
Dunaif A, Graf M, Mandeli J, Laumas V, Dobrjansky A. Characterization of groups of hyperandrogenic women with acanthosis nigricans, impaired glucose tolerance, and/or hyperinsulinemia. J Clin Endocrinol Metab 1987; 65: 499–507. Dahlgren E, Johansson S, Lindstedt G, et al. Women with polycystic ovary syndrome wedge resected in 1956 to 1965: a long-term followup focusing on natural history and circulating hormones. Fertil Steril 1992; 57: 505–13. Wild S, Pierpoint T, McKeigue P, Jacobs H. Cardiovascular disease in women with polycystic ovary syndrome at long-term follow-up: a retrospective cohort study. Clin Endocrinol (Oxf) 2000; 52: 595–600. Franks S, Gharani N, Waterworth D, et al. The genetic basis of polycystic ovary syndrome. Hum Reprod 1997; 12: 2641–48. Goldzieher MW, Green JA. The polycystic ovary I: clinical and histological features. J Clin Endocrinol Metab 1962; 22: 325–38.
22
23
24
25
26
Franks S. Polycystic ovary syndrome: a changing perspective. Clin Endocrinol (Oxf) 1989; 31: 87–120. Balen AH, Conway GS, Kaltsas G, et al. Polycystic ovary syndrome: the spectrum of the disorder in 1741 patients. Hum Reprod 1995; 10: 2107–11. Farquhar CM, Birdsall M, Manning P, Mitchell JM, France JT. The prevalence of polycystic ovaries on ultrasound scanning in a population of randomly selected women. Aust N Z J Obstet Gynaecol 1994; 34: 67–72. Polson DW, Adams J, Wadsworth J, Franks S. Polycystic ovaries—a common finding in normal women. Lancet 1988; 1: 870–72. Zawadzki JK, Dunaif A. Diagnostic criteria for polycystic ovary syndrome: towards a rational approach. In: Dunaif A, Givens JR, Haseltine FP, Merriam GR, eds. Polycystic ovary syndrome. Oxford: Blackwell Scientific Publications, 1992: 59–69. Robinson S, Kiddy D, Gelding SV, et al. The relationship of insulin insensitivity to menstrual pattern in women with hyperandrogenism and polycystic ovaries. Clin Endocrinol (Oxf) 1993; 39: 351–55. Gilling-Smith C, Willis DS, Beard RW, Franks S. Hypersecretion of androstenedione by isolated thecal cells from polycystic ovaries. J Clin Endocrinol Metab 1994; 79: 1158–65. Gilling-Smith C, Story H, Rogers V, Franks S. Evidence for a primary abnormality of thecal cell steroidogenesis in the polycystic ovary syndrome. Clin Endocrinol (Oxf) 1997; 47: 93–99. Willis DS, Watson H, Mason HD, Galea R, Brincat M, Franks S. Premature response to luteinizing hormone of granulosa cells from anovulatory women with polycystic ovary syndrome: relevance to mechanism of anovulation. J Clin Endocrinol Metab 1998; 83: 3984–91. Gougeon A. Regulation of ovarian follicular development in primates: facts and hypotheses. Endocrine Rev 1996; 17: 121–54. Hughesdon PE. Morphology and morphogenesis of the SteinLeventhal ovary and of so-called “hyperthecosis”. Obstet Gynecol Surv 1982; 37: 59–77. Wright CS, Hovatta O, Margara R, et al. Effects of follicle stimulating hormone and serum substitution on the in vitro growth and development of early human ovarian follicles. Hum Reprod 1999; 14: 1555–62. Hovatta O, Silye R, Abir R, Krausz T, Winston RM. Extracellular matrix improves survival of both stored and fresh human primordial and primary ovarian follicles in long-term culture. Hum Reprod 1997; 12: 1032–36. Baker T. A quantitative and cytological study of germ cells in human ovaries. Proc R Soc London (B) 1963; 158: 417–33. Kurilo LF. Oogenesis in antenatal development in man. Hum Genet 1981; 57: 86–92. Melica F, Chiodi S, Cristoforoni PM, Ravera GB. Reductive surgery and ovarian function in the human—can reductive ovarian surgery in reproductive age negatively influence fertility and age at onset of menopause? Int J Fertil Menopausal Stud 1995; 40: 79–85. Lass A, Sile R, Abrams D-C, et al. Follicular density in ovarian biopsy of infertile women: a novel method to assess ovarian reserve. Hum Reprod 1997; 12: 1028–31 Durlinger AL, Gruijters MJ, Kramer P, et al. Anti-Mullerian hormone inhibits initiation of primordial follicle growth in the mouse ovary. Endocrinology 2002; 143: 1076–84. Filho F, Baracat E, Lee TH, et al. Aberrant expression of growth differentiation factor-9 in oocytes of women with polycystic ovary syndrome. J Clin Endocrinol Metab 2002, 87: 1337–44 Vendola K, Zhou J, Wang J, Bondy CA. Androgens promote insulinlike growth factor-I and insulin-like growth factor-I receptor gene expression in the primate ovary. Hum Reprod 1999; 14: 2328–32. Abbott DH, Dumesic D, Franks S. Developmental origin of polycystic ovary syndrome: a hypothesis. J Endocrinol 2002; 174: 1–5.
THE LANCET • Vol 362 • September 27, 2003 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet.
1021
Formation and early development of follicles in the polycystic ovary
L J Webber, S Stubbs, J Stark, G H Trew, R Margara, K Hardy, S Franks
Summary Background Polycystic ovary syndrome is the most common cause of anovulatory infertility. It has long-term health implications and is an important risk factor for type 2 diabetes. However, little is known about the cause of polycystic ovaries. We have used detailed morphological analysis to assess the hypothesis that there is an intrinsic ovarian abnormality that affects the earliest stages of follicular development. Methods We took small cortical biopsies during routine laparoscopy from 24 women with normal ovaries and regular cycles and from 32 women with polycystic ovaries, 16 of whom had regular, ovulatory cycles and 16 of whom had oligomenorrhoea. We used computerised image analysis to assess the density and developmental stage of small preantral follicles in serial sections of fixed tissue. Findings Median density of small preantral follicles, including those at primordial and primary stages, was six-fold greater in biopsies from polycystic ovaries in anovulatory women than in normal ovaries (p=0·009). In both ovulatory and anovulatory women with polycystic ovaries, we noted a significant increase in the percentage of early growing (primary) follicles and a reciprocal decrease in the proportion of primordial follicles compared with normal ovaries. Interpretation Our findings indicate that there are fundamental differences between polycystic and normal ovaries in early follicular development, suggesting an intrinsic ovarian abnormality. The increased density of small preantral follicles in polycystic ovaries could result from increased population of the fetal ovary by germ cells, or from decreased rate of loss of oocytes during late gestation, childhood, and puberty. Lancet 2003; 362: 1017–21
Institute of Reproductive and Developmental Biology, Wolfson and Weston Research Centre for Family Health, Imperial College London, Hammersmith Hospital, London, UK (L J Webber MRCOG, S Stubbs MSc, G H Trew MRCOG, R Margara MD, K Hardy PhD, Prof S Franks FMedSci); and Department of Mathematics, Imperial College London, 180 Queens Gate, London (Prof J Stark PhD) Correspondence to: Prof S Franks, Wolfson and Weston Research Centre for Family Health, Institute of Reproductive and Developmental Biology, Imperial College London, Hammersmith Hospital, London W12 0NN, UK (e-mail: [email protected])
Introduction Polycystic ovary syndrome is the most common endocrine abnormality in women of reproductive age and is the main cause of anovulatory infertility. It is also associated with characteristic metabolic abnormalities, including insulin resistance, that are associated with an increased risk of type 2 diabetes1,2 and cardiovascular disease3 in later life. There are, therefore, important implications for women’s health. The cause of the syndrome remains uncertain; although, there is evidence that genetics have an important role.4 Clinical and biochemical features of polycystic ovary syndrome are typically heterogeneous,5–7 but the distinctive ovarian morphology seen in about a fifth of the female population, is, by most definitions, a characteristic feature of the syndrome.8–10 The major marker of the polycystic ovarian morphology is hyperandrogenaemia,11 and theca cells of polycystic ovaries have been shown to be the principal source of excess androgens.12,13 In anovulatory women with polycystic ovaries, the later (antral) stages of follicular development are clearly abnormal and growth of these follicles is typically arrested at a diameter of 5–8 mm—ie, well before a mature follicle would be expected to ovulate. This aberration of follicular maturation seems to indicate an abnormal endocrine environment,14 but whether there is an intrinsic abnormality of ovarian folliculogenesis remains unclear. Usually, growth of follicles from primordial to preovulatory stages takes 6 months or more15 and only the last 2 weeks of follicle development are highly dependent on cyclical changes in gonadotropins. Earlier in the process, preantral follicle development is primarily dependent on local growth factors and steroids of ovarian origin. These factors determine the survival and growth of a proportion of follicles but most are destined to die by atresia. In ovulatory women, only a few oocytes will eventually reach the preovulatory stage. Little is known about follicle formation and preantral growth in the polycystic ovary.16 We aimed to use detailed morphometric analysis of human ovarian cortical biopsies to test the hypothesis that there is an intrinsic abnormality of folliculogenesis in the polycystic ovary. We predicted that the early stages of follicle development would be abnormal in women with a history of anovulatory polycystic ovary syndrome and in those with polycystic ovaries but regular cycles.
Methods Participants Patients were recruited from the Hammersmith, Queen Charlotte’s and Chelsea, and St Mary’s Hospitals, London, UK, between December, 1997, and September, 2002. We obtained cortical ovarian biopsies from women who were having laparoscopic investigation or surgical treatment for infertility, or other surgical procedures. We established polycystic ovarian morphology or normal
THE LANCET • Vol 362 • September 27, 2003 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet.
1017
ARTICLES
ovarian morphology with ultrasonography. We did ultrasonography, measured mid-luteal progesterone concentrations, or both for evidence of ovulation. Serum testosterone and gonadotropins were measured by specific radioimmunoassay with methods described elsewhere.9 Serum testosterone was not routinely measured in ovulatory participants but we did measure concentrations of this hormone in anovulatory women with polycystic ovaries, We obtained serum in the early follicular phase (days 2–4 of the cycle) to measure gonadotropin concentrations. Cortical tissue was taken from the surface of the ovary opposite the hilum either with a biopsy device (Cook, Hertfordshire, UK) during laparoscopy or with a scalpel during a laparotomy. In women with regular cycles, the biopsy was, in most cases, taken in the mid-luteal phase. If a dominant follicle or corpus luteum was present, we took a sample from the contralateral ovary. Biopsies had a maximum volume of 5 mm3. The study was approved by the research ethics committees of the Imperial College School of Medicine in Hammersmith, Queen Charlotte’s and Chelsea and Acton Hospitals, and the St Mary’s NHS Trust. Patients gave written consent. Tissue preparation and histological assessment Ovarian tissue was gathered into HEPES-buffered minimum essential medium for immediate transfer at 37°C to the laboratory, where it was divided into pieces less than 1 mm in diameter under sterile conditions. One control piece was fixed in Bouin’s solution, stored in alcohol, and later dehydrated and blocked in paraffin wax. Serial sections, 5 m thick, were cut from the blocks, mounted on microscope slides, and stained with haematoxylin and eosin. We prepared about 24 slides, all containing a maximum of seven sections, for each patient. Samples were anonymised and coded so that the observer (LJW) was unable to identify the type of ovary from which the biopsy was taken until data collection was complete. Assessment of follicle development and survival Slides were examined either on a standard Olympus light microscope with a ⫻63 objective or on a Nikon Eclipse E600 light microscope (⫻60 objective) by one observer (LJW). Images were captured with a DXM1200 digital camera (Nikon, Kingston-upon-Thames, UK) with LUCIA image analysis software (Nikon). The number of follicles in each of the tissue pieces was counted and their stage of development and health were assessed. To ensure the most accurate assessment, we examined each follicle in every section on which it appeared. The stage of follicle development was assessed as follows: those with one layer of flattened pregranulosa cells were judged to be primordial; when there was a single layer of cells in which most were expanded cuboidal granulosa cells, the follicle was described as primary. Secondary follicles had more than one layer of granulosa cells, but if there were more than five layers, we classified the follicle as multilayered preantral.17 We judged follicles as healthy or atretic on the basis of morphology. A degenerated oocyte nucleus, uneven or folded nuclear membrane, vacuoles in the oocyte or pyknotic nuclei of several granulosa cells were all regarded as signs of atresia (degeneration) of follicles.18 Atretic follicles could not be staged accurately. Follicle diameter was measured with on-screen calipers provided by the LUCIA software. A second observer (SS) reanalysed a number of sections to confirm accuracy and reproducibility of measurements. Tissue volume was measured to allow the calculation of follicle density—ie, follicle count per mm3 of tissue. The
1018
area of every fifth section of tissue was digitally measured by image analysis (LUCIA image analysis software). We then calculated volume as mean section area⫻number of sections ⫻5 m (ie, section thickness). Statistical analysis We compared ages of women with polycystic ovaries with those of women with normal ovaries using ANOVA: significance was judged to be p⭐0·05. We compared follicle densities in ovarian tissue samples (⭓0·1 mm3) with the Kruskal-Wallis test (non-parametric ANOVA) using the GraphPad InStat version 3.0a for Macintosh (GraphPad Software, San Diego, California USA). If significant, paired comparisons were done with the Mann Whitney test and a Bonferroni correction. Proportions of resting (primordial) and growing follicles in each of the groups (normal, ovulatory polycystic ovaries, and anovulatory polycystic ovaries) were analysed with binomial regression, allowing for over-dispersion. Analysis of proportions is complicated by the fact that different samples had varying numbers of follicles. As a result, estimates of proportions in individual samples varied widely in their accuracy, with samples that had a large number of follicles providing more accurate estimates. For example, in a sample containing only two follicles, the proportion could only be one of 0%, 50%, or 100%. Unless the true value was one of these, the estimate could be substantially inaccurate. Furthermore, even if the true value were close to one of 0%, 50%, or 100% we would still expect to see large errors because of the random effects of sampling. Thus, estimates of proportion, and their comparison between different groups should take into account the number of follicles in each sample. The easiest way to take this variable into account is to assume that each follicle in a sample behaves independently. However, this assumption is unlikely to be true because we can expect correlations between the follicles taken from the same sample, and might result in a phenomenon called over-dispersion—ie, variability in the observed proportions is much greater than that predicted by a binomial distribution. There are several standard ways of addressing this issue. We used one of the most widely used methods, and estimated the dispersion of the observed data and then scaled the SE by this amount. Estimates of proportions, 95% CI and significance levels were all done with the binary regression command in Stata8 (Stata Corporation, College Station TX, USA). This tool calculates a maximum likelihood (or minimum deviance) estimate, computes SE with the resulting information matrix, and then scales them using a deviance-based estimate of the over-dispersion. Role of funding source The sponsors of this study had no role in study design, data collection, data analysis, data interpretation, or the writing of the report.
Results 40% of patients that we asked agreed to participate, and 56 women donated cortical ovarian samples. 32 had polycystic ovarian morphology, and 24 had normal ovaries. All women were having investigation or surgical treatment for infertility, apart from three who were undergoing reversal of sterilisation and one who was having a hysterectomy. All those with normal ovaries and 16 of those with polycystic ovaries had evidence of ovulation, as shown by ultrasonography, normal midluteal progestorone concentrations, or both. The remaining 16 women with polycystic ovaries had
THE LANCET • Vol 362 • September 27, 2003 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet.
ARTICLES
Number of follicles/mm3
100 80 60 40 20 0 O O al rm y PC y PC o N tor tor a a ul ul Ov nov A
O O al rm y PC y PC o N tor tor a a ul ul Ov nov A
O O al rm y PC y PC o N tor tor a a ul ul Ov nov A
Total Primordial Primary Figure 1: Median density of healthy and atretic preantral follicles in cortical biopsies PCO=polycystic ovaries. Biopsy samples were ⭓0·1 mm3. Samples taken from 22 women with normal ovaries, 14 ovulatory women with polycystic ovaries, and 15 anovulatory women with polycystic ovaries. Bars are 95% CI.
100 80 Follicles (%)
oligomenorrhoea or amenorrhoea, were known to be anovulatory, and had clinical or biochemical evidence of hyperandrogenism (ie, they fitted the classic definition of polycystic ovary syndrome10). Mean age did not differ between groups: normal: 32·9 years (SE 1·0); ovulatory polycystic ovaries: 32·4 years (1·1); anovulatory polycystic ovaries: 30·2 (1·4) (p=0·2, ANOVA). Ovarian volume varied significantly between the three groups (p=0·002, ANOVA). Mean volume, calculated from ultrasonograph measurements,9 was greater in anovulatory polycystic ovaries (11·0 mm3 [1·2]) than in either ovulatory polycystic ovaries (7·5 mm3 [0·8]) (p<0·05) or controls (6·0 mm3 [0·5]; p<0·01, posthoc Tukey-Kramer multiple comparisons test). Mean serum testosterone concentrations in anovulatory women with polycystic ovaries was higher than that in a reference group of 50 women with normal ovaries and regular cycles (2·5 nmol/L [0·6] vs 1·4 nmol/L [0·1]; p<0·0001, Student’s unpaired t test). Serum concentrations of luteinising hormone in anovulatory women with polycystic ovaries did not differ from those in ovulatory women with polycystic ovaries and controls (7·3 IU/L [1·2]; 5·7 IU/L [1·0]; and 5·6 IU/L [0·7], respectively). Serum folicle-stimulating hormone concentrations were similar in all three groups (5·6 IU/L [0·3]; 5·5 IU/L [0·6]; and 6·5 IU/L [0·5] IU/L). None of the participants had a raised concentration of serum folicle-stimulating hormone (>12·8 IU/L). More than 90% of follicles in biopsies from both polycystic and normal ovaries were at either the primordial or the primary stage. Secondary follicles were occasionally seen, but only one early antral follicle was noted (from a normal ovary) and there were no multilayered preantral follicles. There were, therefore, too few secondary and preantral follicles to allow accurate calculation of their follicle density. Median follicle diameter was 40·2 m (range 20–105). Median volumes of tissue analysed for the three groups were not significantly different for ovulatory polycystic ovaries, anovulatory polycystic ovaries, or normal ovaries (0·40 mm3 [0·05–0·92], 0·38 [0·09–4·18], and 0·44 mm3 [0·09–3·06], respectively). Follicle densities in tissue from polycystic and normal ovaries are shown in figure 1. Density of the total population of small preantral follicles differed between groups: normal ovaries 11·4 follicles per mm3 (4–34); ovulatory polycystic ovaries 27·4 follicles per mm3 (9–81);
60 40 20 0 al
O
rm
No
at
ul
Ov
y or
PC
at
ul
v no
al
O
PC
y or A
Primordial
O
rm
No
at
ul
Ov
O
PC
y or
PC
at
ul
v no
y or
A
Growing
Figure 2: Mean proportion of total healthy small preantral follicles at the primordial and primary stages PCO=polycystic ovaries. Bars are 95% CI.
anovulatory polycystic ovaries 73·0 follicles per mm3 (31–94) (p=0·009, Kruskal-Wallis). The density of primary follicles also differed significantly between groups (p=0·006, Kruskal-Wallis). When tissue from women with polycystic ovaries was classified by ovulation status, anovulatory polycystic ovaries had a higher overall density of follicles than did normal ovaries (p=0·009, Mann Whitney with Bonferroni correction) but there were no significant differences in density between anovulatory and ovulatory polycystic ovaries (p=0·38) or between ovulatory polycystic ovaries and normal tissue (p=0·4). Primordial follicle density did not differ between the three groups (p=0·3, KruskalWallis) but the difference in density of primary follicles was significant (p=0·006, Kruskal-Wallis) and, as with total follicle density, the major difference was between anovulatory polycystic and normal ovaries (p=0·004, Mann Whitney with Bonferroni correction). We did not note significant differences between the primary follicle densities in normal and ovulatory polycystic ovaries. The proportion of healthy primordial follicles was lower in both types of polycystic ovaries than in normal ovaries (anovulatory polycystic ovaries, p=0·001; ovulatory polycystic ovaries, p=0·03) (figure 2). Compared with normal ovaries, the proportion of healthy primary follicles that had started growing was significantly higher in anovulatory (p=0·001) and ovulatory (p=0·03) polycystic ovaries. We did not note a significant difference in proportions between the two polycystic ovary groups (p=0·7). In normal ovaries, the mean proportion of atretic follicles was 19% (95% CI 14–27), 28% (22–35) in polycystic ovaries from anovulatory women, and 23% (14–35) in ovulatory polycystic ovaries. Although the difference between anovulatory polycystic and normal ovaries approached significance (p=0·06; binomial regression), proportions did not differ between ovulatory polycystic and normal ovaries (p=0·59), or between the two polycystic ovary groups (p=0·38).
Discussion Median density of small preantral follicles, including those at primordial and primary stages, was six-fold greater in anovulatory polycystic ovaries than in normal ovaries. In ovulatory and anovulatory women with polycystic ovaries, the proportion of early growing (primary) follicles was higher than that in normal ovaries, and we noted a reciprocal decrease in the proportion of primordial follicles.
THE LANCET • Vol 362 • September 27, 2003 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet.
1019
ARTICLES
compared with normal ovaries. Indeed, the trend was towards a higher proportion of atretic follicles in anovulatory More PGC More cell More cells polycystic ovaries than in normal ovaries. entering divisions? enclosed by These data suggest that the dynamics of ovary? pregranulosa early follicular development in polycystic cells? ovaries are quite distinct from those in the Increased density normal ovary. Polycystic ovary of small preantral Increased recruitment of follicles from Normal ovary follicles in polycystic the resting pool to the early growing ovaries phase might be expected to encourage Slower rate premature depletion of the stock of of atresia in primordial follicles in polycystic ovaries polycystic and, therefore, to accelerate the onset of ovaries? menopause. However, there is little evidence that menopause in women with polycystic ovaries occurs earlier than in the general population.2,21 The explanation for this apparent absence of effect on age of menopause might be that the increased recruitment from the resting pool is counterbalanced by the higher initial number of follicles in that pool. An alternative (or additional) factor to be considered is that the rate of progression through the later stages of preantral follicle development could be slower in Primordial Mitosis of Follicle Atresia germ cells germ cells formation the polycystic ovary. In many respects, our data are Figure 3: Fate of germ cells in the human fetal ovary and proposed mechanisms to consistent with the findings of explain the increased density of small preantral follicles in the polycystic ovary Hughesdon16 who made a detailed After migration of primordial germ cells into the developing ovary, the population of germ cells histological comparison of sampled increases by mitosis to reach a maximum of seven million at around 20 weeks’ gestation. sections, comprising cortical, subcortical, Follicle formation (the surrounding of primary oocytes by a single layer of somatic, pregranulosa and medullary tissue, from archived cells) might begin as early as the 12th week in utero. The population of germ cells (both naked normal and polycystic ovaries. Although and in primordial follicles) declines steadily from midgestation, by the process of atresia, to reach a level of around two million by birth. our biopsies were confined to cortical tissue, the focus in our study was on early follicle development and the distribution of primordial Infertility resulting from anovulation is one of the most and primary follicles is restricted to the outer 1–2 mm of common features of polycystic ovary syndrome5–7 and the cortex.22 Hughesdon observed that there was some syndrome is the most frequent cause of anovulation. Our data suggest that abnormalities of folliculogenesis that subcortical dislocation of small preantral follicles (ie, lead to anovulation in the polycystic ovary arise at the primordial and primary) but this was more common in earliest stages of follicle development. There are four polycystic ovaries so that, if anything, our measurements possible explanations for the higher overall density of of the density of primordial and primary follicles in small preantral (and specifically of primary) follicles in ovarian cortex would underestimate differences between polycystic ovaries from anovulatory women (figure 3). groups. Hughesdon noted that the number of growing First, the fetal ovary might be endowed with an increased follicles at all stages (including preantral) was increased population of primordial germ cells. The second in polycystic ovaries. However, he reported that the possibility is that the oogonia undergo more mitotic number of primordial follicles per section was much the divisions in the fetal ovary. Third, follicle formation, same in normal and polycystic ovaries in his study. This which begins at about 12 weeks’ gestation20 and involves finding could reflect the fact that in our study we have specifically examined follicle density using serial, rather assembly of somatic cells around the naked oocytes, might than sampled, sections and, in addition, we were able to be enhanced in the polycystic ovary. Fourth, the high rate classify participants by ovulatory status. of loss of germ cells19 (and their surrounding somatic cells) Ample evidence exists to suggest that polycystic ovary that begins in the fetal ovary and continues until the time syndrome has a major genetic cause.4 As yet, the gene (or of the menopause,15 might be reduced in the polycystic ovary. more probably genes) that contribute to the polycystic The increased density of follicles in adult polycystic ovary phenotype remain uncertain. These observations ovaries suggests that, especially in the case of women with lend support to the view that genes involved in early chronic anovulation, such ovaries are characterised by a follicular growth are candidates in the cause of polycystic greater initial pool of follicles. Analysis of the proportions ovary syndrome. Little is known about the molecules that of primordial (resting) and primary (growing) follicles control recruitment and early growth of follicles, but indicates that, irrespective of ovulation status, there might growth factors in the transforming growth factor beta also be increased recruitment (initiation of growth) from (TGF) family such as anti-Mullerian hormone (AMH)23 the resting follicle pool in polycystic ovaries. The and growth differentiation factor-9 (GDF-9)24 have been proportion of atretic follicles in ovulatory or anovulatory implicated and so have androgens.6 Serum concentrations polycystic ovaries did not differ from that in normal of androgens are typically high in women with polycystic ovaries, indicating that the difference in follicle density is ovary syndrome. Treatment of adult Rhesus monkeys not attributable to a reduced rate of atresia in adulthood with androgens has been shown to increase recruitment Number of germ cells
? Higher initial population of primordial follicles
19
20
19
1020
THE LANCET • Vol 362 • September 27, 2003 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet.
ARTICLES
and stimulate growth of preantral follicles.25 The mechanism of androgen action remains unclear, but Vendola and colleagues25 have suggested that the intraovarian insulin-like growth factor (IGF) system is involved; androgens stimulate expression of IGF-1 and its receptor in early growing follicles. Importantly, prenatal exposure to androgens in Rhesus monkies and sheep produces endocrine changes that are very similar to those in polycystic ovary syndrome.26 Androgens might, therefore, play a part in abnormal early folliculogenesis which could, in turn, lead to abnormal function of theca cells in mature follicles, thus setting up a self-propagating cycle of abnormal follicular growth and function.26 Thus, our findings pave the way for further molecular and cellular studies into the origin of this hitherto puzzling syndrome. We have shown abnormalities of folliculogenesis in both ovulatory and anovulatory polycystic ovaries. The disorder is more pronounced in anovulatory polycystic ovaries and might, in part, contribute to the failure of the development of a dominant follicle and subsequent ovulation. Our findings provide more evidence for the central role of an ovarian abnormality in polycystic ovary syndrome. Contributors L Webber did laboratory studies (including histological analyses) as part of her PhD project and was primarily responsible for recruiting patients and obtaining written consent. S Stubbs assisted in histological analyses. G Trew and R Margara supervised recruitment of patients and took biopsy samples. J Stark did statistical analysis. K Hardy and S Franks conceived and supervised the project.
6 7
8
9 10
11
12
13
14
15 16
17
Conflict of interest statement None declared.
18
Acknowledgments This work was funded by Programme Grants from the Medical Research Council (salary for LJW and laboratory expenses) and Wellbeing (salary for SS and laboratory expenses). We thank Catherine S Wright who collected and prepared some of the tissue samples that were used in these analyses.
19 20 21
References 1
2
3
4 5
Dunaif A, Graf M, Mandeli J, Laumas V, Dobrjansky A. Characterization of groups of hyperandrogenic women with acanthosis nigricans, impaired glucose tolerance, and/or hyperinsulinemia. J Clin Endocrinol Metab 1987; 65: 499–507. Dahlgren E, Johansson S, Lindstedt G, et al. Women with polycystic ovary syndrome wedge resected in 1956 to 1965: a long-term followup focusing on natural history and circulating hormones. Fertil Steril 1992; 57: 505–13. Wild S, Pierpoint T, McKeigue P, Jacobs H. Cardiovascular disease in women with polycystic ovary syndrome at long-term follow-up: a retrospective cohort study. Clin Endocrinol (Oxf) 2000; 52: 595–600. Franks S, Gharani N, Waterworth D, et al. The genetic basis of polycystic ovary syndrome. Hum Reprod 1997; 12: 2641–48. Goldzieher MW, Green JA. The polycystic ovary I: clinical and histological features. J Clin Endocrinol Metab 1962; 22: 325–38.
22
23
24
25
26
Franks S. Polycystic ovary syndrome: a changing perspective. Clin Endocrinol (Oxf) 1989; 31: 87–120. Balen AH, Conway GS, Kaltsas G, et al. Polycystic ovary syndrome: the spectrum of the disorder in 1741 patients. Hum Reprod 1995; 10: 2107–11. Farquhar CM, Birdsall M, Manning P, Mitchell JM, France JT. The prevalence of polycystic ovaries on ultrasound scanning in a population of randomly selected women. Aust N Z J Obstet Gynaecol 1994; 34: 67–72. Polson DW, Adams J, Wadsworth J, Franks S. Polycystic ovaries—a common finding in normal women. Lancet 1988; 1: 870–72. Zawadzki JK, Dunaif A. Diagnostic criteria for polycystic ovary syndrome: towards a rational approach. In: Dunaif A, Givens JR, Haseltine FP, Merriam GR, eds. Polycystic ovary syndrome. Oxford: Blackwell Scientific Publications, 1992: 59–69. Robinson S, Kiddy D, Gelding SV, et al. The relationship of insulin insensitivity to menstrual pattern in women with hyperandrogenism and polycystic ovaries. Clin Endocrinol (Oxf) 1993; 39: 351–55. Gilling-Smith C, Willis DS, Beard RW, Franks S. Hypersecretion of androstenedione by isolated thecal cells from polycystic ovaries. J Clin Endocrinol Metab 1994; 79: 1158–65. Gilling-Smith C, Story H, Rogers V, Franks S. Evidence for a primary abnormality of thecal cell steroidogenesis in the polycystic ovary syndrome. Clin Endocrinol (Oxf) 1997; 47: 93–99. Willis DS, Watson H, Mason HD, Galea R, Brincat M, Franks S. Premature response to luteinizing hormone of granulosa cells from anovulatory women with polycystic ovary syndrome: relevance to mechanism of anovulation. J Clin Endocrinol Metab 1998; 83: 3984–91. Gougeon A. Regulation of ovarian follicular development in primates: facts and hypotheses. Endocrine Rev 1996; 17: 121–54. Hughesdon PE. Morphology and morphogenesis of the SteinLeventhal ovary and of so-called “hyperthecosis”. Obstet Gynecol Surv 1982; 37: 59–77. Wright CS, Hovatta O, Margara R, et al. Effects of follicle stimulating hormone and serum substitution on the in vitro growth and development of early human ovarian follicles. Hum Reprod 1999; 14: 1555–62. Hovatta O, Silye R, Abir R, Krausz T, Winston RM. Extracellular matrix improves survival of both stored and fresh human primordial and primary ovarian follicles in long-term culture. Hum Reprod 1997; 12: 1032–36. Baker T. A quantitative and cytological study of germ cells in human ovaries. Proc R Soc London (B) 1963; 158: 417–33. Kurilo LF. Oogenesis in antenatal development in man. Hum Genet 1981; 57: 86–92. Melica F, Chiodi S, Cristoforoni PM, Ravera GB. Reductive surgery and ovarian function in the human—can reductive ovarian surgery in reproductive age negatively influence fertility and age at onset of menopause? Int J Fertil Menopausal Stud 1995; 40: 79–85. Lass A, Sile R, Abrams D-C, et al. Follicular density in ovarian biopsy of infertile women: a novel method to assess ovarian reserve. Hum Reprod 1997; 12: 1028–31 Durlinger AL, Gruijters MJ, Kramer P, et al. Anti-Mullerian hormone inhibits initiation of primordial follicle growth in the mouse ovary. Endocrinology 2002; 143: 1076–84. Filho F, Baracat E, Lee TH, et al. Aberrant expression of growth differentiation factor-9 in oocytes of women with polycystic ovary syndrome. J Clin Endocrinol Metab 2002, 87: 1337–44 Vendola K, Zhou J, Wang J, Bondy CA. Androgens promote insulinlike growth factor-I and insulin-like growth factor-I receptor gene expression in the primate ovary. Hum Reprod 1999; 14: 2328–32. Abbott DH, Dumesic D, Franks S. Developmental origin of polycystic ovary syndrome: a hypothesis. J Endocrinol 2002; 174: 1–5.
THE LANCET • Vol 362 • September 27, 2003 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet.
1021