Twenty-nine clinical features were examined at baseline for identifying PCOS subgroups. The overall value of each feature was assessed through Spearman correlation. Features with a value less than 10% were excluded.2
There were 9 continuous variables included in the final analysis. These included body mass index (BMI), luteinizing hormone (LH), follicle-stimulating hormone (FSH), testosterone, SHBG, dehydroepiandrosterone sulfate (DHEA-S), antimüllerian hormone (AMH), fasting insulin, and fasting glucose.2
Sample collection and laboratory analysis
Investigators collected blood samples at the first consultations. On days 1 to 3 of menstruation, measurements of reproductive steroid hormone levels were also performed. An enzyme-linked immunosorbent assay was used to measure AMH levels and biochemical parameters.2
Participants were diagnosed between 2014 and 2018, then underwent follow-up over the phone between March 2021 and August 2024. A median follow-up duration of 6.5 years was reported, during which patients reported current height, weight, and menstrual cycle status.2
Additional follow-up assessments included pregnancies and births during the follow-up period, IVF during the follow-up period, and disease status. Participants also underwent a physical examination. Among the study population, the following PCOS subtypes were reported:2
- HA-PCOS in 25%
- OB-PCOS in 26%
- SHBG-PCOS in 26%
- LH-PCOS in 23%
Characteristics of PCOS subtypes
The first of these subtypes was characterized by significant DHEA-S levels and mild metabolic disorders, while the second was characterized by an increased BMI and fasting glucose and insulin levels. The third was characterized by the greatest SHBG level and lowest BMI, while the fourth was characterized by increased LH, FSH, and AMH levels.2
Investigators also performed 5 independent validation cohorts that included women from China, the US, Europe, Singapore, and Brazil. The same 4 subtypes were identified in each of these cohorts. Area under the curves for the accuracy of prediction in each cohort were 0.88, 0.92, 0.88, 0.98, and 0.82, respectively.2
In the HA-PCOS subgroup, 67.2% of patients undergoing follow-up still met the Rotterdam criteria for PCOS. In the OB-PCOS, SHBG-PCOS, and LH-PCOS subgroups, these rates were 50.9%, 52.8%, and 74.8%, respectively.2
Implications
These results highlighted the presence of 4 PCOS subtypes with unique characteristics. Investigators concluded these subtypes indicated the potential for more personalized treatment among patients with PCOS.2
“This international collaboration has provided robust evidence that could change how we diagnose, treat, and follow up on women with PCOS. It also emphasizes that treatment can be tailored to the PCOS subtype to better capture the biological variation in PCOS,” said Stener-Victorin.1
References
- Four subgroups of PCOS open up for individualized treatment. News release. Karolinska Institutet. October 29, 2025. Accessed November 3, 2025. https://www.eurekalert.org/news-releases/1103614
- Gao X, Zhao S, Du Y, et al. Data-driven subtypes of polycystic ovary syndrome and their association with clinical outcomes. Nat Med. October 29, 2025. doi.org:10.1038/s41591-025-03984-1