The predictive power of saliva electrolytes exceeds that of saliva microbiomes in diagnosing early childhood caries

Abstract

Early childhood caries (ECC) is one of the most prevalent chronic diseases affecting children worldwide, and thus its etiology, diagnosis, and prognosis are of particular clinical significance. This study aims to test the ability of salivary microbiome and electrolytes in diagnosing ECC, and their interplays within the same population. We here simultaneously profiled salivary microbiome and biochemical components of 331 children ( 166caries-free (H group) and 165 caries-active children (C group)) aged 4-6 years. We identified both salivary microbial and biochemical dysbiosis associated with ECC. Remarkably, K + , Cl -, NH 4 + , Na + , SO 4 2-, Ca 2+ , Mg 2+ , and Brwere enriched while pH and NO 3 were depleted in ECC. Moreover, the dmft index (ECC severity) positively correlated with Cl -, NH 4 + , Ca 2+ , Mg 2+ , Br -, while negatively with pH and NO 3 -. Furthermore, machine-learning classification models were constructed based on these biomarkers from saliva microbiota, or electrolytes (and pH). Unexpectedly, the electrolyte-based classifier (AUROC = 0.94) outperformed microbiome-based (AUROC = 0.70) one and the composite-based one (with both microbial and biochemical data; AUC = 0.89) in predicting ECC. Collectively, these findings indicate ECC-associated alterations and interplays in the oral microbiota, electrolytes and pH, underscoring the necessity of developing diagnostic models with predictors from salivary electrolytes.

Biofilm Accumulation and Sucrose Rinse Modulate the Calcium and Fluoride Bioavailability in the Saliva of Children with Early Childhood Caries

“…(2021) demonstrated that the assessment of children based only on the salivary biochemical electrolytes outperformed the prediction status of children with ECC when compared with classi ers based only on the salivary microbiome or based on both properties of saliva (microbial and biochemical electrolytes data). That is, the assessment of the caries prediction status in children was more reliable when it was based on the assessment of the inorganic composition of saliva (Zhang et al, 2021). Focusing on the bioavailability of the most important ions involved in the demineralization and remineralization events (Ca 2+ , Pi, and F − ), our research brings some light and caveats to this question.…”

Artificial intelligence in salivary biomarker discovery and validation for oral diseases

“…Utilizing AI algorithms with salivary biomarkers to generate models that can be validated and applied potentially in clinical practice is more common than its use for biomarker discovery (Arias-Bujanda et al, 2020;Banavar et al, 2021;Bostanci et al, 2018;Carnielli et al, 2018;da Costa et al, 2022;de Dumast et al, 2018;Eriksson et al, 2022;Gomez Hernandez et al, 2021;Grier et al, 2021;Kim et al, 2021;Kistenev et al, 2018;Koller et al, 2021;Koopaie et al, 2021;Kouznetsova et al, 2021;Lee et al, 2021;Liu, Tong, et al, 2021;Lyashenko et al, 2020;Monedeiro et al, 2021;Nakano et al, 2014Nakano et al, , 2018Pang et al, 2021;Schulte et al, 2020;Shoukri et al, 2019;Song et al, 2020;Sonis et al, 2013;Tamaki et al, 2009;Winck et al, 2015;Wu et al, 2021;Zhang et al, 2021;Zhou et al, 2021;Zlotogorski-Hurvitz et al, 2019) AI-based biomarker platforms constructed during biomarker discovery may be executed for biomarker validation on the premise that each additional saliva sample would be subjected to similar large-scale analysis and used as input for the same models (Adeoye, Wan, et al, 2022). While this may not be cost-effective, feasible, or encourage validation, promising biomarkers selected using conventional statistical approaches or AI-based exploratory analysis (in biomarker discovery) may be used to develop new models for biomarker validation and potential clinical application (Koopaie et al, 2021;Tamaki et al, 2009).…”

“…Utilizing AI algorithms with salivary biomarkers to generate models that can be validated and applied potentially in clinical practice is more common than its use for biomarker discovery (Arias‐Bujanda et al, 2020; Banavar et al, 2021; Bostanci et al, 2018; Carnielli et al, 2018; Chen, Chen, et al, 2021; Chen, Wu, et al, 2021; da Costa et al, 2022; de Dumast et al, 2018; Eriksson et al, 2022; Gomez Hernandez et al, 2021; Grier et al, 2021; Kim et al, 2021; Kistenev et al, 2018; Koller et al, 2021; Koopaie et al, 2021; Kouznetsova et al, 2021; Lee et al, 2021; Li et al, 2021; Liu, Tong, et al, 2021; Lyashenko et al, 2020; Monedeiro et al, 2021; Nakano et al, 2014, 2018; Pang et al, 2021; Schulte et al, 2020; Shoukri et al, 2019; Song et al, 2020; Sonis et al, 2013; Tamaki et al, 2009; Winck et al, 2015; Wu et al, 2021; Zhang et al, 2021; Zhou et al, 2021; Zlotogorski‐Hurvitz et al, 2019). Many of the AI‐assisted salivary biomarker platforms would constitute supervised learning models implemented with structured data since disease or event indicators are often less taxing to annotate (with fewer biomarkers) before model training.…”

Biofilm accumulation and sucrose rinse modulate calcium and fluoride bioavailability in the saliva of children with early childhood caries

“…The most intriguing question here is: Can ECC as a chronic disease influence these electrolytes and explain the boundaries between a healthy and sick oral ecosystem? In this sense, a recent study 13 demonstrated that the assessment of children based only on the salivary biochemical electrolytes outperformed the prediction status of children with ECC when compared with classifiers based only on the salivary microbiome or on both properties of saliva (microbial and biochemical electrolytes data). That is, assessment of the caries prediction status in children was more reliable when based on the inorganic composition of saliva 13 .…”

“…In this sense, a recent study 13 demonstrated that the assessment of children based only on the salivary biochemical electrolytes outperformed the prediction status of children with ECC when compared with classifiers based only on the salivary microbiome or on both properties of saliva (microbial and biochemical electrolytes data). That is, assessment of the caries prediction status in children was more reliable when based on the inorganic composition of saliva 13 . Focusing on the bioavailability of the most important ions involved in demineralization and remineralization events (Ca 2+ , P i , and F − ), we intended to conduct research that brings some light and caveats to this question.…”