Tomás Inmaculada, Regueira-Iglesias Alba, López Maria, Arias-Bujanda Nora, Novoa Lourdes, Balsa-Castro Carlos, Tomás Maria
Oral Sciences Research Group, Department of Surgery and Medical Surgical Specialties, School of Medicine and Dentistry, Universidade de Santiago de Compostela, Health Research Institute of Santiago (IDIS)Santiago de Compostela, Spain.
Department of Microbiology, Complejo Hospitalario Universitario A Coruña-Instituto de Investigación Biomédica de A CoruñaA Coruña, Spain.
Front Microbiol. 2017 Aug 9;8:1443. doi: 10.3389/fmicb.2017.01443. eCollection 2017.
Currently, there is little evidence available on the development of predictive models for the diagnosis or prognosis of chronic periodontitis based on the qPCR quantification of subgingival pathobionts. Our objectives were to: (1) analyze and internally validate pathobiont-based models that could be used to distinguish different periodontal conditions at site-specific level within the same patient with chronic periodontitis; (2) develop nomograms derived from predictive models. Subgingival plaque samples were obtained from control and periodontal sites (probing pocket depth and clinical attachment loss <4 mm and >4 mm, respectively) from 40 patients with moderate-severe generalized chronic periodontitis. The samples were analyzed by qPCR using TaqMan probes and specific primers to determine the concentrations of , , , , , , and . The pathobiont-based models were obtained using multivariate binary logistic regression. The best models were selected according to specified criteria. The discrimination was assessed using receiver operating characteristic curves and numerous classification measures were thus obtained. The nomograms were built based on the best predictive models. Eight bacterial cluster-based models showed an area under the curve (AUC) ≥0.760 and a sensitivity and specificity ≥75.0%. The cluster showed an AUC of 0.773 (sensitivity and specificity = 75.0%). When and were incorporated in the cluster, we detected the two best predictive models with an AUC of 0.788 and 0.789, respectively (sensitivity and specificity = 77.5%). The cluster had an AUC of 0.785 (sensitivity and specificity = 75.0%). When was incorporated in this cluster, a new predictive model appeared with better AUC and specificity values (0.787 and 80.0%, respectively). Distinct clusters formed by species with different etiopathogenic role (belonging to different Socransky's complexes) had a good predictive accuracy for distinguishing a site with periodontal destruction in a periodontal patient. The predictive clusters with the lowest number of bacteria were and , while had the highest number. In all the developed nomograms, high concentrations of these clusters were associated with an increased probability of having a periodontal site in a patient with chronic periodontitis.
目前，基于龈下致病共生菌的定量聚合酶链反应（qPCR）来开发用于慢性牙周炎诊断或预后的预测模型的证据很少。我们的目标是：（1）分析并进行内部验证基于致病共生菌的模型，该模型可用于区分同一慢性牙周炎患者特定部位的不同牙周状况；（2）开发源自预测模型的列线图。从40例中重度广泛性慢性牙周炎患者的对照部位和牙周部位（探诊深度和临床附着丧失分别<4 mm和>4 mm）获取龈下菌斑样本。使用TaqMan探针和特异性引物通过qPCR分析样本，以确定 、 、 、 、 、 和 的浓度。基于致病共生菌的模型通过多变量二元逻辑回归获得。根据指定标准选择最佳模型。使用受试者工作特征曲线评估判别能力，从而获得多种分类指标。基于最佳预测模型构建列线图。八个基于细菌簇的模型显示曲线下面积（AUC）≥0.760，敏感性和特异性≥75.0%。 簇的AUC为0.773（敏感性和特异性=75.0%）。当 和 纳入 簇时，我们检测到两个最佳预测模型，AUC分别为0.788和0.789（敏感性和特异性=77.5%）。 簇的AUC为0.785（敏感性和特异性=75.0%）。当 将 纳入该簇时，出现了一个新的预测模型，其AUC和特异性值更好（分别为0.787和80.0%）。由具有不同致病作用的物种（属于不同的Socransky菌群）形成的不同簇对于区分牙周炎患者中存在牙周破坏的部位具有良好的预测准确性。细菌数量最少的预测簇是 和 ，而 中的细菌数量最多。在所有开发的列线图中，这些簇的高浓度与慢性牙周炎患者出现牙周部位的概率增加相关。
