An explorative study identifies miRNA signatures for the diagnosis of non-celiac wheat sensitivity

Abstract

Non-celiac wheat sensitivity (NCWS), also referred to as non-celiac gluten sensitivity, is a recently described disorder triggered by wheat/gluten ingestion. NCWS elicits a wide range of symptoms including diarrhoea, intestinal discomfort, and fatigue in analogy with other wheat/gluten-related disorders and celiac disease in particular. From the pathological standpoint, NCWS patients only have a slight increase of intraepithelial lymphocytes, while antibodies to tissue transglutaminase (tTG) and villous atrophy, otherwise diagnostic features of celiac disease, are absent. To date, the diagnosis of NCWS relies on symptoms and exclusion of confounding diseases, since biomarkers are not yet available. Here, the expression levels of selected miRNAs were examined in duodenal biopsies and peripheral blood leukocytes collected from newly diagnosed patients with NCWS and, as controls, from patients with celiac disease and gluten-independent gastrointestinal problems. We identified a few miRNAs whose expression is higher in the intestinal mucosa of patients affected by NCWS in comparison to control patients affect by gluten-independent dyspeptic symptoms (Helicobacter pylori-negative) and celiac disease. The present study provided the first evidence that NCWS patients have a characteristic miRNA expression patterns, such peculiarity could be exploited as a biomarker to the diagnosis of this disease.

Fig 1. Volcano plot of the Log2 fold changes versus the -Log10 of p values (Control vs NCWS).

Total RNA extracted from duodenal biopsies of NCWS patients and control patients affect by gluten-independent dyspeptic symptoms was used to PCR amplify 136 selected miRNA. The fold changes of miRNA expression were calculated according to the 2^-ΔΔCt method. Statistical differences between the two groups (control vs NCWS) were calculated using a linear regression model, adjusted for age and gender. Vertical dotted lines indicate thresholds of fold changes. The miRNA above the dashed red line have a p<0.05, while those significantly different after FDR correction are highlighted in red.

Fig 2. Significant miRNAs after FDR correction in duodenal biopsies of the pilot cohort.

The figure reports the ΔCt of miRNA in controls (CTRL) and NCWS patients as well as the fold change (2^-ΔΔCt) with the relative confidence intervals. The p-value relative to a linear model were adjusted for age and gender. ΔCt are expressed as mean±standard deviation (SD). In the right panel, the fold change values and their confidence intervals are shown graphically.

Fig 3. A principal component analysis of miRNA expression levels in duodenal biopsies can classify patients with NCWS and controls.

(A) A principal component analysis was performed on the ΔCt values for each gene. First component (PC1) was plotted against second component (PC2) to identify genes driving the two components and to explain the percentage variation. $•$ Control group; $+$ NCWS group. (B) Fitted ROC curve (blue line) of the first principal component data (red dots). Plotting of the true positive rate versus false positive rate as defined by the PC1 values. AUC = 0.74 ± 0.0702 (p<0.001, Wilcoxon U test). Grey lines: 95% confidence interval of the fitted ROC curve.

Fig 4. A principal component analysis of miRNA expression levels in duodenal biopsies can classify with NCWS and CD.

(A) A principal component analysis was performed on the ΔCt values for each gene. First component (PC1) was plotted against second component (PC2) to identify genes driving the two components and to explain the percentage variation. $+$ NCWS group; $◦$ CD group. (B) Fitted ROC curve (blue line) of the first principal component data (red dots). Plotting of the true positive rate versus false positive rate as defined by the PC1 values. AUC = 0.76 ± 0.0680 (p<0.001, Wilcoxon U test). Grey lines: 95% confidence interval of the fitted ROC curve.

Fig 5. Significant miRNAs in PBL of the validation cohort.

The figure reports the ΔCt of miRNA in controls (CTRL) and NCWS patients as well as the fold change (2^-ΔΔCt) with the relative confidence intervals. The p-value relative to a linear model were adjusted for age and gender. ΔCt are expressed as mean±standard deviation (SD). In the right panel, the fold change values and their confidence intervals are shown graphically.

Fig 6. Principal component analysis of miRNA expression levels in PBL can classify patients with NCWS and controls.

(A) A principal component analysis was performed on the ΔCt values for each gene. First component (PC1) was plotted against second component (PC2) to identify genes driving the two components and to explain the percentage variation. $•$ Control group; $+$ NCWS group. (B) Fitted ROC curve (blue line) of the first principal component data (red dots). Plotting of the true positive rate versus false positive rate as defined by the PC1 values. AUC = 0.84 ± 0.0624 (p<0.001, Wilcoxon U test). Grey lines: 95% confidence interval of the fitted ROC curve.