Diagnostic performance of the systemic inflammation response index and aggregate index of systemic inflammation in polycystic ovary syndrome: A ROC curve analysis
Abstract
Background
Polycystic ovary syndrome (PCOS) is a common endocrine disorder in reproductive-age women, characterized by hormonal, metabolic, and inflammatory changes. Chronic low-grade inflammation plays a key role in its pathogenesis. Traditional markers such as CRP and leukocyte count are often non-specific, while novel biomarkers SIRI and AISI provide a more comprehensive assessment of immune status.
Methods
In this comparative study, 35 women with PCOS and 45 healthy controls aged 18-40 years were evaluated. Age and BMI were comparable between groups. Serum fasting glucose, fasting insulin, lipid profile, CRP, and total testosterone were measured (Cobas pure, Roche Diagnostics), and insulin resistance was estimated via HOMA-IR. SIRI and AISI indices were calculated from complete blood count data using a Sysmex XN-1000 analyzer.
Results
Among women with PCOS, phenotype A was predominant (85.7%). Compared to controls, women with PCOS had higher fasting insulin, HOMA-IR, atherogenic lipid profile, CRP, total leukocyte count, total testosterone, and inflammatory indices (SIRI and AISI), while fasting glucose was similar. Both AISI and SIRI were significantly higher in PCOS (AISI median 269.4 vs. 150.1; SIRI median 0.9500 vs. 0.5600; p = 0.001). ROC analysis showed acceptable discriminatory ability (AUC 0.712 for AISI; 0.713 for SIRI), with moderate sensitivity (AISI 57.8%; SIRI 64.4%) and specificity (AISI 80%; SIRI 74.3%) for distinguishing PCOS from controls.
Conclusions
AISI and SIRI values are significantly higher in women with PCOS and serve as accessible, reliable biomarkers of chronic low-grade inflammation, supporting their use in routine clinical assessment for early detection of systemic inflammation.
© 2026 Natalia Nikolaeva Dikova, Petya Andreeva, Ivanka Dimova, Milena Velizarova, published by Romanian Association of Laboratory Medicine
This work is licensed under the Creative Commons Attribution 4.0 License.