期刊
JOURNAL OF DENTAL RESEARCH
卷 92, 期 12, 页码 1081-1088出版社
SAGE PUBLICATIONS INC
DOI: 10.1177/0022034513506011
关键词
pathogenesis; gene expression; transcriptome; microarray analysis; classification; machine learning
资金
- National Institutes of Health (NIH) [DE-015649, DE-021820, UL1-TR000040, K99 DE-018739]
- Colgate-Palmolive, NJ, USA
- German Research Foundation [KFO208, TP6, TP9]
- German Society for Periodontology (DGParo)
- German Society for Oral and Maxillofacial Sciences (DGZMK)
- Neue Gruppe
- Michael Smith Foundation for Health Research
The 2 major forms of periodontitis, chronic (CP) and aggressive (AgP), do not display sufficiently distinct histopathological characteristics or microbiological/immunological features. We used molecular profiling to explore biological differences between CP and AgP and subsequently carried out supervised classification using machine-learning algorithms including an internal validation. We used whole-genome gene expression profiles from 310 healthy' or diseased' gingival tissue biopsies from 120 systemically healthy non-smokers, 65 with CP and 55 with AgP, each contributing with 2 diseased' gingival papillae (n = 241; with bleeding-on-probing, probing depth 4 mm, and clinical attachment loss 3 mm), and, when available, a healthy' papilla (n = 69; no bleeding-on-probing, probing depth 4 mm, and clinical attachment loss 4 mm). Our analyses revealed limited differences between the gingival tissue transcriptional profiles of AgP and CP, with genes related to immune responses, apoptosis, and signal transduction overexpressed in AgP, and genes related to epithelial integrity and metabolism overexpressed in CP. Different classifying algorithms discriminated CP from AgP with an area under the curve ranging from 0.63 to 0.99. The small differences in gene expression and the highly variable classifier performance suggest limited dissimilarities between established AgP and CP lesions. Future analyses may facilitate the development of a novel, intrinsic' classification of periodontitis based on molecular profiling.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据