Adapted from ref. 14.
studies, including the one in which we compared 214 patients with PCOS to 112 women with normal ovaries (14). By ROC analysis, a follicle number per ovary (FNPO) of 12 or more follicles of 2-9 mm diameter yielded the best compromise between sensitivity (75%) and specificity (99%) for the diagnosis of polycystic ovaries (Table 3). It is not possible to compare these data to the recent modification of their criteria proposed by Adams et al. (9), that is, eight or more cysts 2-8 mm in diameter in a single plane with a peripheral distribution and the impression of increased stroma in the absence of such statistical approach in this last study (14a).
The Rotterdam consensus meeting did not address the difficult issue concerning the presence of multifollicular ovaries (MFO) observed in clinical conditions other than PCOS. Again, these ovaries might be more correctly termed as multifollicular rather than multicystic. There is no consensual definition for the MFO, although they have been described as ovaries in which there are multiple (>6) follicles, usually 4-10 mm in diameter, with normal stromal echogenicity (9). No histological data concerning MFO are available. The MFO are characteristically seen during puberty and in women recovering from hypothalamic amenorrhea—both situations being associated with follicular growth without consistent recruitment of a dominant follicle (15,16). Although the clinical pictures are theoretically different, there may be some overlap—hence the confusion between polycystic ovaries and MFO by inexperienced ultrasonographers. This stresses the need for considering carefully the other clinical and/or biological components of the consensual definition for PCOS. We recently re-examined the ovarian follicular pattern in a group of women with hypothalamic amenorrhea. About one-third had an FNPO higher than 12 (unpublished personal data). Because they were oligo- or anovulatory, they could be considered as having PCOS if one applied too inflexibly the Rotterdam definition! Some of these patients may truly have PCOS, whose clinical and biological expression has been modified by the chronically suppressed LH levels observed in secondary hypothalamic dysfunction (17). In others, however, such an overlap in the FNPO emphasizes the need for a wise and careful utilization of the Rotterdam criteria as well as for considering other ultrasound criteria for polycystic ovaries in difficult situations.
In the 1970s, the weak resolution of U/S abdominal probes only allowed the detection exclusively of external morphological ovarian features, which were used as the first criteria for defining polycystic ovaries:
1. An ovarian length greater than 4 cm was a simple criterion. However, this unidimensional approach may lead to many false-positive results, particularly when a full bladder compresses the ovary (with the transabdominal route), or false-negative results when the ovaries are spherical with a relatively short length.
However, we should note that these parameters are less used today because of their low sensitivity and specificity (18).
The ovarian area is less often used than ovarian volume and was not retained in the consensus definition, but in our recent study the diagnostic value of the ovarian area (assessed by ROC curves) was slightly better than that of the ovarian volume (sensitivity 77.6%; specificity 94.7% for a threshold at 5 cm2/ovary) (13). We also observed that the measured ovarian area (obtained after outlining the ovary by hand) was more informative than the calculated ovarian area (by using the formula for an ellipse: L x W x W x rc/4). Indeed, ovaries are not strictly ellipsoid. We previously reported that the sum of the area of both ovaries was less than 11 cm2 in a large group of normal women (19,20), but a threshold of 5 cm2 per ovary seems to offer the best compromise between sensitivity and specificity. Consequently, if the ovarian area (for a single ovary) is >5 cm2, the diagnosis of polycystic ovaries is suggested (13).
Stromal hypertrophy is frequently present in PCOS and is characterized by an increased central ovarian component, which is rather hyperechoic (Fig. 1). In others (21) and our (18,19) opinion, stromal hypertrophy and hyperechogenicity are helpful in distinguishing between polycystic ovaries and MFO, because these features are specific for the former. However, the estimation of hyperechogenicity is considered highly subjective, mainly because it depends on the settings of the U/S machine. Likewise, in the absence of a precise quantification, stromal hypertrophy is a relatively subjective sign.
For standardizing the assessment of stromal hypertrophy, we designed a computerized quantification of ovarian stroma, allowing for selective calculation of the stromal area by subtraction of the cyst area from the total ovarian area on a longitudinal ovarian cut (19,20). By this means we were able to set the upper normal limit of the stromal area (i.e., 95th percentile of a control group of 48 normal women) at 380 mm2 per ovary. However, providing a precise outlining of the ovarian shape on a strictly longitudinal cut of the ovaries, the diagnostic value of the total ovarian area equaled that of stromal area, as both were highly correlated.
Fulghesu et al. (22) proposed the ratio of the ovarian stroma-to-total area as a reliable criterion for the diagnosis of PCOS. The ovarian stromal area was evaluated by outlining the peripheral profile of the stroma with the caliper, identified by a central area slightly hyperechoic with respect to the other ovarian area. However, this evaluation is not easy to reproduce in routine practice.
Stromal echogenicity has been described by others (23) in a semi-quantitative manner, where a score was assigned for normal (1), moderately increased (2), or frankly increased (3) stroma. In this study the total follicle number of both ovaries combined correlated significantly with stromal echogenicity. Stromal echogenicity has also been quantified by Al-Took et al. (24) as the sum of the product of each intensity level (ranging from 0 to 63 on the scanner) and the number of pixels for that intensity level divided by the total number of pixels in the measured area. Buckett et al. (25) used this same formula, but found no difference of the stromal echogenicity between women with PCOS and those with normal ovaries. Their conclusion was that the subjective impression of increased stromal echogenicity is a result of both increased stromal volume and reduced echogenicity of the multiple surrounding follicles.
In summary, because ovarian volume or area correlate well with ovarian function and it is easier to reliably measure these than ovarian stroma in routine practice, neither qualitative nor quantitative assessment of the ovarian stroma is required to define polycystic ovaries.
In polycystic ovaries the follicular distribution is predominantly peripheral, typically with an echoless peripheral array, as initially described by Adams et al. (9) (Fig. 1). Younger patients more often display this peripheral distribution, whereas a more generalized pattern, with small cysts in the central part of the ovary, is noticed in older women (26). At the Rotterdam meeting, this subjective criterion was judged too inconstant and subjective to be retained for the consensus definition of polycystic ovaries (1).
2.5. Other U/S Techniques for Imaging Polycystic Ovaries 2.5.1. 3D Ultrasound
To avoid the difficulties and pitfalls in outlining or measuring the ovarian shape, the use of 3D U/S has been proposed (27-29). Using a dedicated volumetric probe and a manual survey of the ovary, the scanned ovarian volume is displayed on the screen in three adjustable orthogonal planes. This allows the three ovarian dimensions, and subsequently the ovarian volume, to be more accurately calculated. In a study by Kyei-Mensah et al. (30), three groups of patients were defined: (1) those with normal ovaries, (2) those with asymptomatic polycystic ovaries, and (3) those with PCOS. The ovarian and stromal volumes were similar in groups 2 and 3, and both had greater volumes than group 1. Stromal volume was positively correlated with serum androstenedione levels in group 3 only. The mean total volume of the follicles was similar in all groups, indicating that increased stromal volume is the principal cause of ovarian enlargement in polycystic ovaries. Nardo et al. (31) observed a tight correlation between 2D and 3D ultrasound measurements of ovarian volume and polycystic ovary morphology. However, in this prospective study total ovarian volume, ovarian stromal volume, fol-licular volume, and follicle numbers did not correlate with testosterone concentration.
Because 3D ultrasound requires expensive equipment, intensive training, and significant storage and data analysis, its superiority over 2D ultrasound for imaging polycystic ovaries in clinical practice is not evident.
Color (or power) Doppler U/S allows for detection of the vascularity network within the ovarian stroma. Power Doppler is more sensitive to slow blood flow and consequently demonstrates more vascular signals within the ovaries, although it does not discriminate between arteries and veins. Moreover, the sensitivity of the machines currently marketed differs from one to another. The pulsed Doppler focuses on the hilum or internal ovarian arteries and offers a more objective approach. Because of the slow blood flow in the ovary, the pulse repetition frequency should be set at minimum (~400 Hz) with the lowest frequency filter (~50 Hz).
The study of ovarian vascularity by these techniques is still highly subjective. Blood flow is more frequently visualized in PCOS (88%) than in normal patients (50%) in early follicular phase and seems to be greater in the former (32). No significant difference was found between obese and lean women with polycystic ovaries, but the stroma was less vascularized in patients displaying a general cystic pattern than in those with peripheral cysts. In the latter group, the pulsatility index (PI) values were significantly lower and inversely correlated with the FSH-to-LH ratio (33). In another study (34), the resistive index (RI) and PI were significantly lower in PCOS (RI = 0.55 ± 0.01 and PI = 0.89 ± 0.04) than in normal patients (RI = 0.78 ± 0.06 and PI = 1.87 ± 0.38), and the peak systolic velocity was greater in PCOS (11.9 ± 3.2) than in normal women (9.6 ± 2.1). No correlation was found between the number of follicles and the ovarian volume, although there was a positive correlation between LH levels and increased peak systolic velocity. In a study by Zaidi et al. (35), no significant differences in PI values were found between normal and PCOS patients, while the ovarian flow, as reflected by the peak systolic velocity, was increased in the former. It is possible that Doppler-de-tected blood flow may have some value in predicting the risk for ovarian hyperstimulation during gonadotropin therapy (36). Increased stromal blood flow has also been suggested as a more relevant predictor of ovarian response to ovulation induction compared to other parameters such as ovarian or stromal volume (25,37).
In summary, with ovarian Doppler U/S the increased stromal component in polycystic ovaries appears to be accompanied by an increased peak systolic velocity and a decreased PI. However, in all studies the values in patients with polycystic ovaries overlapped widely with those of normal women, and there are few data to date to support the diagnostic usefulness of Doppler U/S in polycystic ovaries or PCOS.
Was this article helpful?