Making the diagnosis of PCOS remains a difficult challenge and a controversial issue, especially in adolescent girls. In particular, whether ultrasonography should be used as a universal standard is still disputed.

Hormonal Markers for PCOS
Before the development of ultrasonography (see below), the diagnosis of PCOS was for many years mainly based on an elevated serum LH level and/or an exaggerated LH response to the GnRH test and/or an elevated LH/FSH ratio. However, in other conditions clearly different from PCOS such as mild hypothalamic anovulation or hyperprolactinemia, the LH level may be elevated after GnRH stimulation. Therefore, many authors have questioned the sensitivity as well as the specificity of a high LH level. Recent data emphasized that obesity is a modifier of the biological expression of PCOS in that it attenuates the excessive LH plasma level, by reducing the LH pulse amplitude but not its frequency.

In the basal state, the biological hyperandrogenism of PCOS cannot define the syndrome. In most cases, A and T serum levels are modestly and inconstantly elevated, in a non-specific manner. For some authors, non-suppression of serum-free T after DXM defines functional ovarian hyperandrogenism in cases where other well-characterized causes of hyperandrogenism have been excluded.

More recently, the GnRH agonist test was proposed as a new functional approach to define PCOS. The dissociation between the 17-HP and A responses suggested a dysregulation of the cytochrome P₄₅₀ C17α. Therefore, the GnRH agonist test appears to provide an integrated probe to detect the thecal dysregulation of PCOS, in the face of either normal or abnormal LH serum levels. The results of this test correlated well with some other features of PCOS, in particular the serum-free T response to the dexamethasone (DXM) suppression test and the ovarian stromal area. The ovarian stimulation HCG test yielded similar results. However, they are not convenient for the diagnosis of PCOS in practice.

Morphological and Ultrasound Criteria
Numerous parameters have been proposed to define PCO by ultrasound but there is still no consensus about their use for the diagnosis of PCOS. They can be classified as either external or internal morphologic features. Features that describe directly or indirectly the ovarian hypertrophy are relatively objective since they are obtained by measuring physical entities (e.g., ovarian axis, area or volume) in real-time conditions. The ovarian area was previously considered as normal up to 10cm² while the threshold is now almost two times lower (5.5cm²) in a more recent series compared to a large control group.

The internal morphologic features describe the two main histologic features of PCO, namely the increased number of subcapsular follicles and the stromal hyperplasia. Their analysis is more subjective, especially by the abdominal route, which is the rule in adolescents (

fig. 2). Furthermore, the follicular pattern (number and size of follicles) is a non-specific sign which can also be observed in hyperprolactinemia, functional hypothalamic anovulation or even in late puberty or in the early normal follicular phase. Conversely, the hyperechogenicity of ovarian stroma is considered as suggestive of PCO, but it is highly dependent on the settings of the ultrasound machine. In order to avoid the subjectivity of this sign in clinical practice, it is recommended that the total ovarian area be used since it is highly correlated to the stromal area. The presence of an ovarian hypertrophy at ultrasonography certifies PCO, but its absence does not exclude the diagnosis.

Screening for Metabolic Syndrome
This should include clinical parameters (family history, BMI, WHR, blood pressure, search for acanthosis nigricans) and biological parameters (glycemia, insulinemia, HDL-C, triglycerides). The optimal means for the assessment of hyperinsulism is still under debate. Insulin levels are often routinely used as an index of insulin resistance but they also reflect both pancreatic secretory capacity and insulin extraction by the liver. Glucose clamp techniques and mathematical modeling to measure insulin sensitivity are not easily applied to clinical practice. Indirect markers of insulin action may be more valuable tools. The SHBG level is highly and negatively correlated to the insulin level.  Likewise, the IGFBP-1 level is strongly suppressed by hyperinsulinism.