The role of ultrasound and Doppler analysis in the diagnosis of polycystic ovary syndrome.

Abstract

PCOS, in its classic form, is characterized by infertility, oligomenorrhea and/or amenorrhea, hirsutism, acne or seborrhea, and obesity. In 1935 the syndrome was recognized by Stein and Leventhal2 in a group of seven hirsute, amenorrheic women on the basis of typical ovarian morphology: fibrotic thickening of the tunica albuginea and outer cortex, and multiple cystic follicles with prominent thecae. The incidence of the syndrome varies considerably between different reports. In 1951 Vara and Niemineva3 in a series of 12 160 unselected gynecological laparotomies found polycystic ovaries (PCO) in 1.4% of patients. Sommers and Wadman4, a few years later, diagnosed in 740 autopsies the typical ovarian expression of PCOS in 3.5% of the cases. Clayton et al.5 more recently claimed that the finding of PCO is more common, with 22% of the younger female population showing a PCO-link pattern on ultrasound scanning of the ovaries. Adams et al.6, using transabdominal ultrasound, found PCO in 26% of patients with amenorrhea, in 87% of those with oligomenorrhea and in 92% of women with hirsutism. Furthermore, using an epidemiological approach, Hull7 calculated a possible incidence of PCOS of 90% in infertile oligomenorrheic patients, and of 37% in infertile amenorrheic patients. From these various studies one can conclude that the occurrence of PCO is relatively common in the female population8 and that there is a correlation between ovarian appearance, cycle history and clinical evidence of excess androgen. PCOS may result from disturbances of various endocrine systems. It has been diagnosed in patients with adult onset of adrenal 21-hydroxylase deficiency, Cushing's syndrome, adrenal hyperplasia, hypo- and hyperthyroidism, adrenal or ovarian tumors, and hyperprolactinemia9-15. However, excluding these conditions, and despite a large amount of epidemiological, clinical, laboratory and experimental studies, the etiology and pathophysiology of the syndrome still remain fragmentary and obscure, and should probably be considered multifactorial16. Increased pre- or peripubertal adrenal function may result in elevated circulating androgens, which, by peripheral conversion (in fat and/or brain tissue), are changed into estrogens. The excessive estrogen levels are responsible for a disturbed (pulse frequency and/or amplitude) pattern of luteinizing hormone (LH) secretion. This, in turn, stimulates the ovarian thecal compartment to secrete androgens17. When this vicious circle is established, it is possible to identify the typical structural changes in the ovary. Hyperandrogenism and hyperinsulinemia were linked as early as 1921, when Achard and Thiers18 published their study on bearded women with diabetes. Since then, researchers have realized that insulin resistance may be implicated in the pathophysiology of PCOS19. Furthermore, Adashi et al.20 found that insulin may potentiate LH secretion in PCOS patients. It is well known that hyperinsulinemia may lead to increased ovarian androgen production and to enhanced conversion of testosterone into 5α-dihydrotestosterone. In addition, it has been shown that hyperandrogenism may result from increased activity of cytochrome P-450c17-α, and that it may inhibit the liver's production of sex-hormone-binding globulin, resulting in a further increase of circulating androgens21. Finally, insulin-like growth factor-1 (IGF-1) increases the expression of LH receptors and stimulates LH-induced androgen production and its ovarian accumulation. Although no significant differences in IGF-1 have been observed in PCOS vs. non-PCOS patients, biological IGF-1 activity may be increased by a decrease of the specific binding globulin22. Poretsky and Piper23 recently postulated that elevated LH and hyperinsulinemia, acting synergistically, induce ovarian stromal and thecal hyperplasia, hyperandrogenism and follicular atresia. The subsequent predominance of androgen-secreting cells may be responsible for PCOS clinical manifestations. It seems that both the elevated LH secretion and the insulin resistance of PCOS are genetically determined. The insulin receptor gene and the β-LH subunit gene have been mapped to chromosome 1923-25. Clinically, a strong association between maternal PCOS and sonographic PCO in prepubertal offspring has been observed26, supporting the hypothesis of genetic transmission of the syndrome. It has also been postulated that PCO in childhood may be considered a sign of genetic predisposition to PCOS and that environmental factors (mainly nutritional) may lead to expression in the adult, causing clinical and biochemical presentation of the syndrome27. Although some familial studies on PCOS have identified both X-linked and autosomal dominance, other segregation analyses have found an incidence of affected relatives exceeding that which would be expected from autosomal dominance28-30. These inconsistencies when trying to determine the familial contribution are probably correlated with the heterogeneity of the syndrome and with the diagnostic criteria used to identify probands with PCOS, as well as to characterize other affected family members. PCOS and the correlated androgen excess may be particularly distressing and disruptive for women. Thus, PCOS requires prompt diagnosis and appropriate treatment. An inadequate diagnostic approach may result in missed diagnosis, whereas an excessive number of tests may increase the likelihood of diagnostic confusion, and favoring the performance of unnecessary tests may lead to economic wastage31. The criteria for diagnosis and definition of PCOS are as heterogeneous as the pathology itself. The predominantly North American definition of PCOS coincides with the view expressed at the 1990 National Institute of Health's conference on PCOS32, which recommended that diagnostic criteria should include biochemical evidence of hyperandrogenism and ovarian dysfunction in the absence of other endocrine disorders. The European definition of PCOS requires the sonographic appearance of PCO associated with menstrual disturbances and clinical signs of hyperandrogenism. No hormonal parameters are required for the definitive diagnosis33. Probably, a streamlined evaluation including history, clinical, hormonal and sonographic evaluation may reduce the risk of under- or over-evaluation of the syndrome. A detailed history concerning the chronobiological development of symptoms is mandatory in establishing the true diagnosis. PCOS patients generally report a history of peripubertal onset of increased hair growth, oligo- and/or amenorrhea, and, in almost 50% of the cases, obesity26. Adequate estrogen activity is, however, confirmed by breast development and secondary signs of sexual maturation. The clinical manifestations of PCOS, as analyzed by Goldzieher and Axelrod34, vary enormously and complicate the diagnosis. The symptoms with the highest average incidence are: hirsutism (69%), infertility (74%), and menstrual disorders (79%). The complexity of the pathophysiological interactions and the heterogeneity of the clinical expression mean that there is no specific hormonal test that can be used for the diagnosis of PCOS. In fact, the endocrinological findings generally correlated with PCOS (elevated LH or LH/follicle stimulating hormone (FSH) ratio, abnormally high androstenedione and/or testosterone levels), are often inconsistent. This discrepancy may be due to the fact that LH and steroid hormones oscillate in pulses of relatively high frequency and are subject to circadian rhythm35. Furthermore, plasma testosterone levels are in part determined by peripheral androstenedione conversion and by the amount bound to both the circulating sex-hormone-binding globulin and albumin9. Elevated plasma estrogen levels are also of limited diagnostic value for classifying PCOS. It should be remembered that about 25% of PCOS patients ovulate and that plasma estrone and estradiol levels are governed by the ovary and/or adrenal gland and by peripheral androstenedione conversion9. However, Koskinen et al.36 showed that the simultaneous use of different analytes (LH, FSH, and androstenedione) may be useful and cost-effective in the diagnosis of PCOS. Nevertheless, the typical appearance of PCO at laparoscopy, whether or not associated with ovarian biopsy, is still considered by many clinicians to be a ‘condicio sine qua non’ for a correct diagnosis37, 38. The advent of sonographic examination of the ovaries has provided the biggest single contribution to the diagnosis of PCOS. This non-invasive technique, performed by a skilled operator, has a high concordance rate with laparoscopy and histological examination and, in the presence of hyperandrogenism and menstrual disorders, may be considered the ‘gold standard’ for the diagnosis of the syndrome. The histoanatomical characteristics of PCOS are variable even though an increased number of follicles and increased ovarian stromal tissue may be considered constant features that can be identified with pelvic ultrasound39. The possibility of studying by ultrasound the female pelvic structures, namely the uterus and ovaries, was first demonstrated in the early 1970s by Kratochwil et al.40, who described the sonographic uterine and ovarian changes in relation to the menstrual cycle. Since then, although sonography has emerged as an essential method for monitoring utero-ovarian activity in many clinical conditions, no complete consensus has been reached on the criteria necessary for PCOS diagnosis, nor on the validity of the various parameters41. In 1985 Adams et al.42 defined the criteria for sonographic diagnosis of PCO as: multiple (n > 10), small (2–8 mm) peripheral cysts around a dense core of stroma in enlarged (≥8 mL) ovaries. However, ovaries which are normal in volume can be polycystic, as demonstrated by histological and biochemical studies. Transabdominal echography has revealed enlarged ovaries (8–14 mL) in up to 70% of symptomatic patients and has demonstrated that the follicles (usually peripherally displaced) may also be scattered throughout the hyperechogenic stroma43, 44. The above findings were based on transabdominal sonography, which, though associated with good specificity as confirmed in patients who underwent ovarian wedge resection45, showed variable results in terms of follicular number, ovarian size, and ovarian stromal echodensity46. It has been claimed that the anatomical structure of the ovaries cannot adequately be assessed using the transabdominal approach in about 42% of cases47. Underlying causes include obesity, limited resolution of low-frequency transducers, a full bladder distorting the pelvic anatomy, and bowel loops covering the adjacent ovary48, 49. More recently, the transvaginal approach for ultrasound scanning of pelvic organs has been introduced. The high frequency of the transvaginal probe avoids the need for a full bladder and bypasses the problems of attenuation and artifacts associated with obesity. Improved resolution leads to better visualization of the pelvic organs, and the transvaginal approach is generally well-tolerated by most patients48-53. More precise sonographic criteria for the diagnosis of PCO can be achieved as findings on transvaginal sonography reflect more closely the histological parameters54, 55. The number of follicles that have to be present to establish the diagnosis of PCO by sonography has been variously reported as ‘more than 5’, ‘more than 10’, and ‘at least 15’42, 56-58. However, in many reports the highest number of atretic follicles obtained in normal control patients was five per ovary, so it may conventionally be established that in the PCO the number of atretic follicles per ovary would be ‘at least 6’. Matsunaga et al.59 identified two types of PCO on the basis of their sonographic follicular distribution: peripheral cystic pattern (PCP) and general cystic pattern (GCP). In the PCP, small cysts are distributed in the subcapsular region of the ovary, whereas in the GCP they are scattered throughout the entire ovarian parenchyma. Recently, Takahashi et al.54 showed that these two different ovarian morphologies reflect histopathological differences, and that the PCP and GCP appearances reflect specific endocrine patterns of PCOS. Another parameter considered in the diagnosis of PCOS is the ovarian volume. However, the wide volume overlap between normal and PCOS patients suggests that the discriminative capacity of ovarian volume alone is not sufficient for ultrasound diagnosis of PCOS. The role of a hyperechogenic ovarian stroma in the differentiation between normal ovary and PCO has been emphasized and a sensitivity of 94% has been reported49. However, the assessment of ovarian stromal echodensity60 is entirely subjective and may be interpreted differently by different operators. Buckett et al.61, using the textural feature analysis, found that the mean stromal echogenicity is not higher in women affected by PCOS compared with women with normal ovaries and suggested that the subjective impression of highly echogenic stroma may be due simply to an increased stromal volume. An ovarian stromal hypertrophy associated with hyperandrogenism and ovulation disorders was reported by Robert et al.62 in 1995. The authors, by using computer-assisted analysis for selective measurements of the stromal area, reported that in PCOS patients the mean stromal area is > 95th percentile of normal controls and that the sign has a specificity of 96%. The advent of other sophisticated new systems such as three-dimensional (3D) ultrasound scanning allows good precision and a high degree of reproducibility of volume measurements. However, although Kyei-Mensah et al.63 found a positive correlation between 3D stromal volume and androstenedione, Nardo et al.64 demonstrated that the 3D measurements of ovarian stromal volume are not associated with biochemical indices of PCOS. The above-mentioned methods of study of PCOS patients are difficult to apply in routine general practice, but recently, the stromal/total ovarian area ratio has been proposed as a criterion for the sonographic evaluation of PCO65. The new criterion seems to be easy to reproduce (interobserver variation <5%) and correlates with testosterone, androstenedione and 17α-hydroxyprogesterone levels and with the free androgen index. Usually, the sonographic modifications are present in both ovaries and most investigators have stressed this bilaterality44, 66, 67 both at laparoscopy37 and histologically38, 54, 55. However, some cases of unilateral PCO have been described68-71. In addition, after the first evidence in 1981 of one case of partial manifestation of PCO68, a further fifteen cases of partial PCO have been described recently72. The authors speculated that in unilateral or in partial PCOS the peripheral expression of ovarian receptors for LH or environmental factors (surgery, infections, ovarian vascularization) may limit the morphological and functional features of PCOS to a single ovary or to a specific portion of the ovaries71, 72. From the above it is evident that the great biological variability is expressed by an equally wide range of echographic patterns and that it is not possible to obtain a cut-off level for any of the analyzed parameters even in the presence of a good sensitivity and specificity. Although Amer et al.73 found that the sonographic criteria currently used for the diagnosis of PCO are associated with such a high intra- and interobserver error that they should be considered subjective criteria, it is likely that a specific diagnosis may be obtained by the combined analysis of all the studied parameters74. The presence of PCO on ultrasound, however, does not automatically confer a diagnosis of PCOS. In fact, whereas some women with functional ovarian hyperandrogenism do not have PCO demonstrable on ultrasound75, 16–25% of normal ovulatory women5, 8 and 27–39% of teenagers76 with no signs of hyperandrogenism show a sonographic pattern of PCO. Thus, it is apparently unclear whether the sonographic presence of PCO represents a variant of normal ovarian appearance or a cryptic or unexpressed form of PCOS. Michelmore et al.76 showed that women with sonographic evidence of PCO have a higher incidence of irregular menstrual cycles than do those with normal ovaries. Carmina et al.77 demonstrated that up to 33% of normally ovulating women with PCO may have increased androgen secretion and a decreased production of insulin-like growth factor binding protein-1, which may be the result of mild insulin resistance. In addition, it has been shown that ovulating women with PCO are more likely to suffer miscarriage78 and subfertility79. These patients have a longer follicular phase, larger follicles, and a higher median follicular concentration of LH, FSH and testosterone compared with ovulating women with normal ovaries80, which might result in the release of aging oocytes81. Furthermore, during controlled ovarian hyperstimulation, apparently eumenorrheic women with PCO respond to gonadotropin with multiple follicular development and a sudden increase in the concentration of circulating estradiol22. In addition, the weight gain in eumenorrheic women with a normal body mass index and the sonographic evidence of PCO is frequently associated with hyperandrogenism and the typical development of PCOS symptoms82. Finally, as mentioned above, it has been postulated that PCO in childhood may be a sign of genetic predisposition to PCOS and that environmental factors (mainly nutritional) may lead to expression in the adult, causing clinical and biochemical presentation of the syndrome27. Thus, the sonographic appearance of PCO should not, in fact, be regarded as a morphological variant of normal ovaries but as a cryptic or unexpressed form of PCOS. Therefore, the diagnosis of PCO should initiate investigation of all the environmental factors involved in the development of the syndrome, and a strict follow-up is mandatory so that early intervention to reduce the reproductive, metabolic and cardiovascular risks linked with PCOS can be undertaken. The introduction of transvaginal Doppler sonography has contributed markedly to the refinement of ultrasound diagnosis. It has also provided much new morphological and pathophysiological information on blood flow dynamics within the female pelvis83-87. The vessels most often analyzed in reproductive endocrinology are the uterine and ovarian arteries. Color flow images of the ascending branches of the uterine arteries lateral to the cervix in a longitudinal plane can be obtained both transvaginally and transabdominally. The ovarian arteries can be evaluated at the level of the ovarian hilum83, 88-94. Recently, attention has been paid to small vessels supplying the ovarian stroma95, 96. It has been shown that in patients with PCOS, important changes in ovarian vascularization occur at the level of the intraovarian arteries. Although intraovarian arteries are usually not seen before day 8–10 of the 28-day cycle95, Battaglia et al.97 detected distinct arteries with characteristic low vascular impedance as early as cycle day 3–5. In the studied population the results were associated with typical PCOS hormonal parameters and were inversely correlated with the LH/FSH ratio. Tonic hypersecretion of LH during the follicular phase of the menstrual cycle occurs in PCOS and is associated with thecal cell and stromal hyperplasia with consequent androgen overproduction. Elevated LH levels may be responsible for increased stromal vascularization through different mechanisms that may act individually or in a cumulative way: neoangiogenesis; catecholaminergic stimulation; leukocyte and cytokine activation98-100. In the same study97, the PCOS patients showed higher uterine artery pulsatility index (PI) values compared with non-hirsute normally menstruating women. This finding was correlated with androstenedione levels confirming a possible direct androgen vasoconstrictive effect due to activation of specific receptors present in the arterial blood vessel wall and collagen and elastin deposition in smooth muscle cells101, 102. The vasoconstrictive role of androgens has been recently confirmed by Ajossa et al.103 who showed that the use of flutamide, a pure non-steroidal antiandrogen, induces, probably by decreasing the circulating dehydroandrostenedione-sulfate levels, an increase in uterine perfusion. Increased resistance in the uterine arteries and subsequent reduction in uterine perfusion was thought to be the cause of failure of the blastocyst to implant and the increased miscarriage rate in PCOS patients. Similar results were obtained by Zaidi et al.104 and Aleem and Predanic105, confirming that Doppler analysis of stromal arteries in PCOS may be useful to improve diagnosis and to provide further information about the pathophysiology and evolution of the syndrome. These hypotheses have been recently confirmed using two-dimensional (2D)106, 107 and 3D108, 109 color Doppler systems. In contrast with Takahashi et al.54, who reported that the GCP appearance is associated with ovarian steroid disorder whereas PCP is associated with abnormal gonadotropin secretion, Battaglia et al.110 suggest that the different ovarian morphologies do not reflect different endocrine patterns but may be considered an evolution of the same endocrinological alteration. In fact, Doppler evaluation110 demonstrated that patients with PCP, in comparison with those with GCP, present significantly lower ovarian stromal artery resistance index (RI) values, and in 22% of patients with GCP the intraovarian vessels were not recognized. In addition, the GCP appearance of the ovary is more common in the early phase of the disease26, 43, 54 during the peripubertal period. Thus, the ovarian morphology may evolve from a normal multicystic to a polycystic PCP pattern passing through an ovarian GCP aspect, and untreated PCOS may be regarded as a progressive syndrome. This has been confirmed recently by Avvad et al.111 and by a longitudinal study on girls with isolated premature pubarche26. Furthermore, in a comparison of oligomenorrheic with amenorrheic PCOS patients, it has been shown that amenorrheic patients are older and present higher uterine artery PI values and lower intraovarian artery RI values112. These findings were associated with higher plasma LH and androstenedione levels and with a more elevated LH/FSH ratio. In addition, significantly greater ovarian volumes and subcapsular small-sized follicles were observed in amenorrheic PCOS patients. These data show that as the number of ovarian microcysts increases, ovarian volume enlarges and Doppler indices worsen, the clinical and endocrine abnormalities become more marked and the menstrual disturbances more severe. Probably, early therapeutic intervention would not only temporarily alleviate symptoms, but may delay the progression of the syndrome, thus reducing the reproductive, metabolic and cardiovascular risks19. It is well known that the assessment of ovarian morphological and functional changes during the treatment of PCOS has been generally lacking or contradictory results have been reported. Genazzani et al.113 demonstrated that the administration of a gonadotropin releasing hormone-analog plus an oral contraceptive is effective in improving PCOS symptoms. Both clinical (Ferriman-Gallwey score) and hormonal (androgens, LH/FSH ratio) parameters improved during the therapy and follow-up period. Subsequently, the same group114 observed that the changes in hormonal findings were associated with complimentary ultrasound (decreased ovarian volume and stromal density, and reduced number of small-sized subcapsular follicles) and Doppler (disappearance of stromal vascularization and improvement of uterine artery PI values) modifications. The above results showed that it is possible, probably by expanding the duration of the therapy, to reach the main ‘goal’ of therapy in PCOS, i.e. to reduce the elevated plasma androgen levels and regulate the associated sequelae. Furthermore, the improved uterine hemodynamic status may contribute to the increased chances of pregnancy in those patients who ovulated normally after therapy. In the same study the changes of arterial tone were not associated with significant variations in lipid profile and body-weight values, confirming the existence of two different PCOS populations: lean and obese PCOS patients. It has been demonstrated that obese PCOS women show higher PI values within the uterine arteries than do lean patients115. This was associated with higher hematocrit values, hyperinsulinemia, higher triglyceride levels and lower high-density lipoprotein (HDL) concentrations. Furthermore, in the obese patients, an inverse correlation was observed between HDL and uterine artery resistance. Insulin resistance is considered to be a risk factor for coronary heart disease, as it is often associated with diabetes mellitus, hypertension and adverse lipid profile. In addition, Wild et al.116 showed that in women undergoing coronary angiography, hirsute patients were more likely to have significant atheroma compared with non-hirsute women. In overweight patients, hyperinsulinemia may be the uniting factor between increased vascular resistance, obesity, lipid abnormalities and cardiovascular diseases. Thus, assuming that PCOS patients are at increased risk for cardiovascular disease, it is possible to affirm that obesity may further increase the risk. In this case transvaginal color Doppler, by evaluating vascular blood flow modifications, may be considered another surrogate marker for possible cardiovascular disease. Despite the complexity of the pathophysiological interactions and the heterogeneity of the anatomical and clinical expression of PCOS, the assessment of ovarian morphology by transvaginal ultrasound and Doppler flow analysis of intraovarian and uterine arteries seems to provide an insight into the pathological status and the degree of progression of the disease.