炎症生物标志物对感音神经性听力损失合并2型糖尿病患者的预测价值

文献信息

中文摘要

目的： 评估炎症标志物（中性粒细胞与淋巴细胞比值[NLR]和血小板与淋巴细胞比值[PLR]）联合临床特征对突发性感音神经性听力损失（SSNHL）合并2型糖尿病（T2DM）患者听力预后的预测价值。
方法： 回顾性分析2018年至2025年间接受治疗的188例SSNHL合并T2DM患者队列。患者被随机分为训练队列（n=132）和测试队列（n=56）。采用最小绝对收缩和选择算子（LASSO）回归联合多因素logistic回归识别听力恢复不良的独立预测因素。构建联合预测模型和可视化列线图。使用曲线下面积（AUC）衡量模型性能，并进行内部验证。
结果： 识别出六个独立预测因素：年龄、从发病到治疗的时间、眩晕、初始纯音平均听阈（PTA）、PLR和NLR。联合模型在训练队列中的AUC为0.899，在测试队列中为0.892，显著优于仅基于临床特征的模型（AUC=0.820）。列线图显示出良好的区分度和校准度。
结论： 将NLR和PLR纳入临床风险模型可显著提高对SSNHL合并T2DM患者听力预后不良的预测。该列线图为个性化风险评估提供了一种用户友好、可床旁使用的工具。

英文摘要

OBJECTIVES: To evaluate the predictive value of inflammatory markers (neutrophil-to-lymphocyte ratio [NLR] and platelet-to-lymphocyte ratio [PLR]) combined with clinical characteristics for hearing prognosis in patients with sudden sensorineural hearing loss (SSNHL) and type 2 diabetes mellitus (T2DM).
METHODS: A retrospective cohort of 188 SSNHL patients with T2DM treated between 2018 and 2025 was analyzed. Patients were randomly divided into a training cohort (n=132) and a test cohort (n=56). Least absolute shrinkage and selection operator (LASSO) regression combined with multivariate logistic regression was used to identify independent predictors for poor hearing recovery. A combined predictive model and a visual nomogram were built. Model performance was measured using the area under the curve (AUC), with internal validation.
RESULTS: Six independent predictors were identified: age, time from onset to treatment, vertigo, initial pure-tone average (PTA), PLR, and NLR. The combined model achieved an AUC of 0.899 in the training cohort and 0.892 in the test cohort, significantly outperforming the model based on clinical features alone (AUC=0.820). The nomogram demonstrated good discrimination and calibration.
CONCLUSIONS: Incorporating NLR and PLR into a clinical risk model substantially improves prediction of poor hearing prognosis in SSNHL patients with T2DM. The nomogram furnishes a user-friendly, bedside-ready means for personalized risk appraisal.