Abstract
To compare the analytical performance of the Roche Cobas Pro and Siemens Atellica Solution platforms for measuring serum luteinizing hormone (LH), follicle-stimulating hormone (FSH), estradiol, and testosterone in pediatric samples, with an emphasis on analytical bias, agreement, and clinical interpretability. A method comparison study was conducted using paired serum samples from children (LH: n = 132; FSH: n = 140; estradiol: n = 413; testosterone: n = 125). An additional 125 adult female samples were included to extend the estradiol measurement range. Hormone concentrations were measured on both platforms according to CLSI EP09-A3 guidelines. Analytical agreement and bias were evaluated using Passing–Bablok regression, Bland–Altman analysis, and Spearman correlation. Strong correlations were observed for all analytes (Spearman r ≥ 0.899; p < 0.0001). However, clinically relevant biases were detected. Compared with Cobas Pro, Atellica underestimated LH (slope 0.856; intercept −0.424 IU/L) and overestimated FSH (slope 1.087). Estradiol demonstrated the greatest disagreement, with Atellica showing higher values in pediatric samples (mean bias +24.65 pmol/L) and adult female samples (mean bias +142.18 pmol/L). Among pediatric estradiol measurements, 232/413 samples were below the limit of quantification (LOQ) on one or both platforms; notably, 147 samples were undetectable by Cobas Pro but quantifiable on Atellica. Testosterone showed near-proportional agreement (slope 0.99), although non-linearity was observed at lower concentrations. Although both platforms demonstrate strong analytical correlation, measurement biases, particularly for LH and estradiol, are of sufficient magnitude to influence clinical interpretation in pediatric endocrinology, including puberty assessment and longitudinal hormone monitoring.