Next-generation sequencing identifies micro-RNA–based biomarker panel for Kawasaki disease

Abstract

Capsule Summary 22Diagnosing Kawasaki disease (KD) currently relies on physicians' experience, 23 making it difficult to accurately diagnose KD. Using NGS, qPCR and machine 24 learning algorithm, we identified a miRNA-based biomarker panel with high 25 accuracy. 26 27 Kawasaki disease (KD) is an autoimmune disease preferentially attacking 32 children younger than five years old. Symptoms of KD include fever for at least 33 five days, oral mucosal inflammation, non-supportive conjunctivitis, 34 lymphadenopathy, edema of the extremities and a polymorphous rash. If left 35 untreated, almost 20-25% of the KD patients may develop severe coronary artery 36 aneurysms, making KD the leading cause of acquired heart disease in young 37 children in developed countries 1 . The successful detection of KD within the first 38 10 days of fever onset followed by treatment with high dose intravenous 39 immunoglobulin (IVIG) can greatly reduce the risk of severe coronary artery 40 lesions 1-3 . Therefore, early and successful detection is critical for treating KD. 41 So far, few cytokine-based biomarker platform was developed 4 . Therefore, the 42 detection of KD still mainly relies on the judgment of clinical physicians based on 43 patients' symptoms. However, infectious diseases by pathogens, may also result in 44 similar symptoms to those of KD, making it difficult to accurately diagnose KD and 45 leading to a delay in optimal and timely treatment of KD. 46 Previous studies showed that it is applicable to use blood miRNAs as disease 47 biomarkers 5-7 . So, we enrolled 70 fever control (FC, denoting non-KD patients with 48 fever) and 50 KD subjects (see Supplementary Methods). Among them, 37 FC and 49 31 KD subjects comprised the training set, used for developing KD biomarker panel 50 with NGS and/or qPCR assays. The remaining 33 FC and 19 KD subjects comprised 51 the blind test set, used for examining the performance of the developed KD 52biomarker panel. Table 1 shows that the most prevalent diagnosis of FC subjects was 53 upper or lower respiratory infection (28 and 27 patients, respectively). In KD 54 patients, bilateral conjunctivitis, oral mucosa changes and rash were the most 55 prevalent clinical features. 56We extracted RNA samples from white blood cells (see Supplementary 57 Methods). Then, we had two pooled FC and two pooled KD RNA libraries 58 sequenced with Illumina NGS platform, followed by miRSeq 8 analysis for 59 determining miRNA expression profiles (Supplementary Table 1). According to 60 miRSeq result, almost 80% of the initial reads belonged to clean read, the sequence 61 read with 3' adaptor recognized and trimmed. In addition, the majority of the clean 62 reads was 22-nt long (Supplementary Figure 1), reflecting high performance of the 63 sample preparation procedures and implying the dominance of miRNAs, which was 64 proven by Supplementary Figure 2. 65 To examine whether the FC and KD samples were distinguishable based on 66 miRNA expression profiles, we conducted clustering analysis. The heat map...

HAMP promoter hypomethylation and increased hepcidin levels as biomarkers for Kawasaki disease

“…Support vector machine (SVM, lib-SVM version 3.22) is a type of machine learning algorithm and is good in dealing with binary classification problem, applied in many biomedical studies [23,28,29]. In this study, we applied the lib-svm module to derive a classification model.…”

“…As a kind of machine learning algorithm, SVM is usually applied to handle biomedical problems to discriminate treatment from control or disease from normal [28,29]. We have also applied SVM to develop a classification model that discriminates febrile controls from KD subjects [23,24]. Since we have identified cg23677000, cg0408544, and hepcidin as potential biomarkers, we used the methylation percentages of cg23677000 and cg04085447, as well as the plasma ELISA level of hepcidin, as SVM vectors for each subject.…”

“…An in vitro reporter gene assay was carried out to validate the transcriptional repression of HAMP. Furthermore, using the CpG sites of HAMP promoter and plasma hepcidin levels, we built a support vector machine (SVM) classification model [23,24] for KD biomarkers.…”

Cognitive development of children with Kawasaki Disease and the parenting stress of their caregivers

“…A senior clinician diagnosed patients with KD in accordance with the recommended universal KD criteria published by the American Heart Association (AHA) [27]. The diagnostic criteria of KD includes fever that lasts longer than ve days, as well as four of the following ve symptoms: diffuse mucosal in ammation with strawberry tongue and ssure lips, bilateral non-purulent conjunctivitis, indurative angioedema of the hands and feet, dysmorphic skin rashes, and unilateral cervical lymphadenopathy, as stated in our previous reports [28]. Once their symptoms were remitted, KD patients were administered a developmental or cognitive assessment in either outpatient department or in the ward.…”

Association between the miRNA‐149 rs2292832 T>C polymorphism and Kawasaki disease susceptibility in a southern Chinese population

“…More than 25 miRNAs have been identified for differentiating patients with KD from patients with fever using next‐generation sequencing. Quantitative polymerase chain reaction was selected for further analysis of some of these miRNAs, but researchers did not find a connection between miRNA‐149 and KD . In searching the literature, we found that miRNA‐149 has a definite association with some cancers, such as lung cancer, colorectal cancer, laryngeal squamous cell carcinoma, hepatocellular carcinoma, and neuroblastoma .…”