Delrue Charlotte, Belpaire Arno, Delanghe Sigurd, Oyaert Matthijs, De Bruyne Sander, Speeckaert Marijn M, Speeckaert Reinhart
Department of Nephrology, Ghent University Hospital, 9000 Ghent, Belgium.
Department of Dermatology, Ghent University Hospital, 9000 Ghent, Belgium.
Diagnostics (Basel). 2025 May 29;15(11):1369. doi: 10.3390/diagnostics15111369.
: Serum diagnostic tests for alopecia areata may be used to monitor response to treatment, aiding in the objective assessment of disease activity and helping to change treatment at an earlier point. Attenuated total reflection Fourier transform infrared (ATR-FTIR) spectroscopy offers a nondestructive and user-friendly approach for analyzing a wide range of samples. In this study, we evaluated whether ATR-FTIR spectroscopy combined with machine learning can detect alopecia areata and quantify disease activity. We also established whether patient-specific spectral differences correlate with response to therapy, offering molecular insight into treatment response. : Serum samples from 42 patients with alopecia areata and 41 healthy donors were compared. Logistic regression models were developed to separate alopecia areata patients from controls and to monitor treatment response based on clinical scoring. : Significant spectral variations were found in the 3000-2800 cm and 1800-1000 cm regions corresponding to the principal biochemical constituents such as proteins, lipids, carbohydrates, and nucleic acids. The AUC of the logistic regression model for distinguishing alopecia areata patients from healthy controls was 0.85 (95% CI: 0.75-0.94) with a sensitivity of 0.89 and a specificity of 0.71. In terms of prediction of treatment response, the model showed discriminative potential (AUC = 0.86, 95% CI: 0.71-0.98), with distinct alterations in the spectrum, particularly in the Amide I band, associated with improvement in the patient's condition. : ATR-FTIR spectroscopy assisted by machine learning offers a serum-based solution for treatment monitoring in alopecia areata patients with clinical applicability. This technique has highly promising potential for the development of rapid, non-invasive, and objective biomarkers in autoimmune dermatology. Additional multi-center trials are required to validate and incorporate these spectral biomarkers into individual treatment regimens.
斑秃的血清诊断测试可用于监测治疗反应，有助于对疾病活动进行客观评估，并有助于在更早阶段改变治疗方案。衰减全反射傅里叶变换红外（ATR-FTIR）光谱法为分析各种样品提供了一种无损且用户友好的方法。在本研究中，我们评估了ATR-FTIR光谱法结合机器学习是否能够检测斑秃并量化疾病活动。我们还确定了患者特异性光谱差异是否与治疗反应相关，从而为治疗反应提供分子层面的见解。：比较了42例斑秃患者和41名健康供体的血清样本。建立了逻辑回归模型，以区分斑秃患者和对照组，并基于临床评分监测治疗反应。：在3000-2800 cm和1800-1000 cm区域发现了与蛋白质、脂质、碳水化合物和核酸等主要生化成分相对应的显著光谱变化。区分斑秃患者和健康对照的逻辑回归模型的AUC为0.85（95%CI：0.75-0.94），灵敏度为0.89，特异性为0.71。在治疗反应预测方面，该模型显示出判别潜力（AUC = 0.86，95%CI：0.71-0.98），光谱有明显变化，特别是在酰胺I带，与患者病情改善相关。：机器学习辅助的ATR-FTIR光谱法为斑秃患者的治疗监测提供了一种基于血清的解决方案，具有临床适用性。该技术在自身免疫性皮肤病学中开发快速、非侵入性和客观的生物标志物方面具有极具前景的潜力。需要进行更多的多中心试验来验证这些光谱生物标志物并将其纳入个体治疗方案。
