鼻腔分泌物多组学揭示伴鼻息肉的慢性鼻窦炎无创内型分型的候选标志特征

文献信息

中文摘要

背景： 伴鼻息肉的慢性鼻窦炎（CRSwNP）在生物学上具有异质性，但用于区分嗜酸性CRSwNP（eCRSwNP）与非嗜酸性CRSwNP（neCRSwNP）的实用无创标志物仍然有限。我们在一个小型试点队列中研究了与组织定义内型相关的鼻腔分泌物蛋白质组学和代谢组学特征。
方法： 采用四维数据非依赖采集（4D-DIA）蛋白质组学和非靶向代谢组学分析对照组（n = 5）、eCRSwNP（n = 5）和neCRSwNP（n = 5）的鼻腔分泌物。采用交集-排除策略在三个两两比较中优先筛选eCRSwNP相关特征。候选蛋白进一步根据表观受试者工作特征（ROC）性能、效应量、探索性多重检验支持和基于重复留一交叉验证（LOOCV）的惩罚回归进行排序。在独立的同中心重复验证集（n = 9）中，用酶联免疫吸附试验（ELISA）测定三种蛋白。在代谢组学质量控制后保留的9对匹配eCRSwNP/neCRSwNP样本中，进行了使用潜在成分的 biomarker 发现数据整合分析（DIABLO）。
结果： 在5874种蛋白质中，4609种通过过滤，全局主成分分析显示组间分离有限。筛选策略优先选出94个蛋白质特征和28个推定注释的代谢特征。基于肿瘤蛋白D52（TPD52）、嗜酸性粒细胞过氧化物酶（EPX）和跨膜蛋白263（TMEM263）的评分在eCRSwNP中高于neCRSwNP和对照组，且在ELISA集中具有一致的方向性趋势。DIABLO保留了这三种蛋白以及推定注释为鸟苷3',5'-环磷酸和UDP-D-半乳糖的特征。在9对匹配样本中，整合模型产生的表观ROC曲线下面积（AUC）为0.95，LOOCV误差为0.167；这些估计值仍然容易受到选择偏倚和过拟合的影响。
结论： 在这项小型单中心研究中，鼻腔分泌物多组学识别出与组织定义eCRSwNP相关的分子特征。这些发现是初步的，需要在更大的独立队列中使用靶向定量和配对组织水平生物学验证加以确认。

英文摘要

BACKGROUND: Chronic rhinosinusitis with nasal polyps (CRSwNP) is biologically heterogeneous, yet practical noninvasive markers that distinguish eosinophilic CRSwNP (eCRSwNP) from non-eosinophilic CRSwNP (neCRSwNP) remain limited. We investigated nasal-secretion proteomic and metabolic features associated with tissue-defined endotypes in a small pilot cohort.
METHODS: Nasal secretions from controls (n = 5), eCRSwNP (n = 5), and neCRSwNP (n = 5) were analyzed by four-dimensional data-independent acquisition (4D-DIA) proteomics and untargeted metabolomics. An intersection-exclusion strategy was used to prioritize eCRSwNP-associated features across three pairwise comparisons. Candidate proteins were further ranked using apparent receiver operating characteristic (ROC) performance, effect size, exploratory multiple-testing support, and repeated leave-one-out cross-validation (LOOCV)-based penalized regression. Three proteins were measured by enzyme-linked immunosorbent assay (ELISA) in a separate same-center replication set (n = 9). Data Integration Analysis for Biomarker discovery using Latent cOmponents (DIABLO) was performed in nine matched eCRSwNP/neCRSwNP samples retained after metabolomic quality control.
RESULTS: Of 5874 proteins, 4609 passed filtering, and global principal component analysis showed limited separation among groups. The screening strategy prioritized 94 protein features and 28 putatively annotated metabolic features. A score based on tumor protein D52 (TPD52), eosinophil peroxidase (EPX), and transmembrane protein 263 (TMEM263) was higher in eCRSwNP than in neCRSwNP and controls, with concordant directional trends in the ELISA set. DIABLO retained these three proteins together with features putatively annotated as guanosine 3',5'-cyclic monophosphate and UDP-D-galactose. In the nine matched samples, the integrated model yielded an apparent area under the ROC curve (AUC) of 0.95 and a LOOCV error of 0.167; these estimates remain vulnerable to selection bias and overfitting.
CONCLUSION: In this small, single-center study, nasal-secretion multi-omics identified molecular features associated with tissue-defined eCRSwNP. These findings are preliminary and require confirmation in larger independent cohorts using targeted quantification and paired tissue-level biological validation.