Adolescents diagnosed with polycystic ovary syndrome under the Rotterdam criteria but not meeting the diagnosis under the updated guideline.
Level 4 - case-series / case-control
Retrospective observational comparative study
PubMed 38514450 · doi:10.1093/humrep/deae042
What was done
A retrospective study (2004–2022) evaluated 315 adolescent girls (2 to 8 years post-menarche) initially diagnosed with polycystic ovary syndrome (PCOS) using the 2003 Rotterdam criteria. Patients were re-stratified according to updated international guidelines—which advise against pelvic ultrasound for diagnosis in girls <8 years post-menarche—and compared against healthy controls.
What was found
Of the 315 girls diagnosed by Rotterdam criteria, phenotypes were: irregular menstruation (IM)/hyperandrogenism (HA)/polycystic ovarian morphology (PCO) in 206 (65.4%), IM/HA in 30 (9.5%), HA/PCO in 12 (3.8%), and IM/PCO in 67 (21.3%). Under the updated guideline, 79 girls (25.1% with HA/PCO or IM/PCO) did not meet formal PCOS criteria and were designated 'at-risk'. Girls meeting updated criteria exhibited the worst metabolic profiles (generalized/central obesity, insulin resistance, dysglycemia, metabolic syndrome) and highest hirsutism scores. Although ~90% of the at-risk group were non-obese (comparable to controls), they had higher blood pressure, triglycerides, insulin resistance parameters, serum LH levels, and LH/FSH ratios than controls, sharing key hormonal and metabolic traits with adolescents meeting full PCOS criteria.
Why it matters
Applying updated pediatric guidelines reclassifies a quarter of Rotterdam-diagnosed adolescents into an 'at-risk' category. While avoiding overdiagnosis via ultrasound, these at-risk girls still display significant cardiometabolic and endocrine disturbances requiring proactive monitoring.
Limits
The study is retrospective across a long timeframe (2004–2022). The abstract does not specify the control group sample size, nor were hormonal and ultrasound parameters evaluated in the healthy controls. Exact numerical estimates, confidence intervals, and p-values were omitted from the abstract.