Total Testosterone Measured by Liquid Chromatograph-Tandem Mass Spectrometry Refines Diagnosis of Biochemical Hyperandrogenism and Better Identifies Subgroup at Genuine Risk of Adverse Fertility Outcomes in Women With Polycystic Ovary Syndrome

Abstract

Purpose: To compare total testosterone (TT) measured by liquid chromatograph-tandem mass spectrometry (LC-MS/MS) with electro-chemiluminescent immunoassay (ECLIA) in the diagnosis and management of infertile women with polycystic ovary syndrome (PCOS).

Methods: Baseline TT was measured by LC-MS/MS and ECLIA in 906 infertile women with PCOS. The associations of TT from both methods with clinical phenotypes and fertility outcomes were estimated; relative risk (RR) and 95% confidence intervals (CIs) were computed. Subgroup analysis was conducted according to the TT levels.

Results: The average TT levels measured by ECLIA were higher than those measured by LC-MS/MS (mean percentage difference 23.8%, 95% limits of agreement -44.2% to 91.9%). When biochemical hyperandrogenism (HA) defined as TT ≥ 1.7 nmol/L by LC-MS/MS method, a higher proportion of patients were identified having biochemical HA using ECLIA (44.0% vs. 24.0%, p < 0.001) than LC-MS/MS. Only those with TT levels ≥ 1.7 nmol/L measured by LC-MS/MS had an increased risk of adverse fertility outcomes compared to patients with normal TT levels, including ovulation, preterm labor, and neonatal intensive care unit.

Conclusion: Our findings indicated that LC-MS/MS refined the diagnosis of biochemical hyperandrogenism and better identified the subgroup at genuine risk of adverse fertility outcomes in infertile women with PCOS.

Trial registration: The NIH Clinical Trial Registry number: NCT01573858 and Chinese Clinical Trial. Registry number: ChiCTR-TRC-12002081.

Keywords: biochemical hyperandrogenism; liquid chromatograph‐tandem mass spectrometry; polycystic ovary syndrome; total testosterone.

FIGURE 1

Passing‐Bablok regression and Bland–Altman graphs. (A) Passing‐Bablok regression graph with depiction of the 95% confidence interval (CI) for both the slope and intercept. Regression line equation (solid red lines): Y = 0.26 + 1.04 x. The 95% CI for intercept spans from 0.18 to −0.33, while for the slope, it ranges from 0.98 to 1.10. These intervals suggest a good agreement between the datasets. (B) The Bland–Altman plot illustrates the mean percentage difference (sold blue line) and the 95% limits of agreement (dashed light green lines). The plot indicates that the mean percentage difference (bias) was 23.8% (95% limits of agreement −44.2% to 91.9%) between ECLIA and LC–MS/MS. ECLIA, electro‐chemiluminescent immunoassay; LC–MS/MS, liquid chromatograph‐tandem mass spectrometry.

FIGURE 2

Assessment of TT, FT, and FAI across different clinical phenotypes. The violin plot showed the distribution of assay‐specific total testosterone (TT), free testosterone (FT), and free androgen index (FAI) across various clinical phenotypes, including hirsutism, polycystic ovaries, overweight/obesity, and low sex hormone‐binding globulin (SHBG) levels. The central thick bar denotes the interquartile range, while the white dot signifies the median value. Hirsutism was defined as modified Ferriman‐Gallwey score ≥ 5. Polycystic ovaries were diagnosed by transvaginal ultrasound when at least one ovary had a volume of > 10 mL or there were 12 or more follicles measuring 2–9 mm in diameter. Overweight/obesity was defined as BMI of ≥ 24 kg/m². Low SHBG levels were set as below 35 nmol/L.

FIGURE 3

Correlation and comparison between assay‐specific TT measurement and other androgen index. (A) Comparison of biochemical hyperandrogenism determined by assay‐specific TT levels. (B) Correlation matrix of assay‐specific TT measurement and associated FAI, modified Ferriman‐Gallwey (mF‐G) score, and clinical/biochemical hyperandrogenism.