Assessment of salivary microRNA by RT-qPCR: Facing challenges in data interpretation for clinical diagnosis

Abstract

Salivary microRNAs (miRNAs) have been recently revealed as the next generation of non-invasive biomarkers for the diagnostics of diverse diseases. However, their short and highly homologous sequences make their quantification by RT-qPCR technique highly heterogeneous and study dependent, thus limiting their implementation for clinical applications. In this study, we evaluated the use of a widely used commercial RT-qPCR kit for quantification of salivary miRNAs for clinical diagnostics. Saliva from ten healthy volunteers were sampled four times within a three month time course and submitted for small RNA extraction followed by RT-qPCR analysed. Six miRNAs with different sequence homologies were analysed. Sensitivity and specificity of the tested miRNA assays were corroborated using synthetic miRNAs to evaluate the reliability of all tested assays. Significant variabilities in expression profiles of six miRNAs from ten healthy participants were revealed, yet the poor specificity of the assays offered insufficient performance to associate these differences to biological context. Indeed, as the limit of quantification (LOQ) concentrations are from 2-4 logs higher than that of the limit of detection (LOD) ones, the majority of the analysis for salivary miRNAs felt outside the quantification region. Most importantly, a remarkable number of crosstalk reactions exhibiting considerable OFF target signal intensities was detected, indicating their poor specificity and limited reliability. However, the spike-in of synthetic target miRNA increased the capacity to discriminate endogenous salivary miRNA at the LOQ concentrations from those that were significantly lower. Our results demonstrate that comparative analyses for salivary miRNA expression profiles by this commercial RT-qPCR kit are most likely associated to technical limitations rather than to biological differences. While further technological breakthroughs are still required to overcome discrepancies, standardization of rigorous sample handling and experimental design according to technical parameters of each assay plays a crucial role in reducing data inconsistencies across studies.

Fig 1. Physical properties and small RNA extraction of saliva samples from 10 healthy participants.

(A) Cesia and (B) viscosity of the saliva samples prior RNA extraction. (C) Total small RNA extracted from 250 μL of saliva for each participant and for the 4 sampling points. Black points represent the mean value and error bars show the standard error of the mean. (D) Detection of spiked artificial UniSP6 miRNA prior to RNA extraction for sampling three. 1 μL of each extracted RNA sample was used as input. P1-10: Saliva sample from participants 1 to 10.

Fig 2. RT-qPCR shows high variability of salivary miRNA expression profiles within 10 healthy participants.

(A) Average Ct values of the four sampling points through which the six miRNAs were assessed on 10 participants. Expression profiles of the six analysed miRNAs across (B) time and (C) different participants. All analysis was perform with 50 ng of total extracted small RNA. All values represent the mean value and the error bars show the standard error of the mean.

Fig 3. Sensitivity and specificity limitations of the six miRNA assays.

(A) Serial dilution from 1 to 10¹² copies/μL with synthetic miRNA target for each miRNA assay to determine their limit of detection (LOD) and their limit of quantification (LOQ). Values represent the mean value and the error bars depict the standard error of the mean. Cross reactions between miRNA assays and synthetic targets at (B) 10⁹ copies/μL and (C) 10⁵ copies/μL. ΔCt values are calculate from S14 Fig. Empty spaces represent no crosstalk (no detection or Ct Value = 35). Data obtained from a duplicate experiment. ΔCt values = CtOFF target–Ct^{ON target}.

Fig 4. Conserved crosstalk for hsa-Let-7a-5p and hsa-Let-7f-5p within the quantification region.

ΔCt values obtained when comparing a serial dilution from 10⁻¹ to 10¹² copies/μL of the corresponding synthetic miRNA (ON target) with its cross reaction (OFF target, red circles) for (A) hsa-Let-7a-5p assay and (B) hsa-Let-7f-5p assay. The ΔCt by the addition of 10⁷ copies/μL of OFF target on the serial dilution of ON target was also calculated (blue triangles). Red circles = Ct^{OFF target}–Ct^{ON target}, blue triangles = Ct^{ON target + OFF target at 10^7 copies/μL}–Ct^{ON target}. All values represent the mean value and the error bars show the standard error of the mean (two independent experiments). Grey region delimits the quantification region.

Fig 5. The spiking of synthetic hsa-Let-7a-5p miRNA allows semi-quantification at LOQ concentrations.

(A) ΔCt values of spike-in synthetic hsa-Let-7a-5p miRNA in 50 ng RNA extract with respect to the RNA extract alone in a range of spike-in concentrations. ΔCt were calculated as CtEndogenous—Ct^{Endogenous + synthetic}. (B) ΔCt values of spike-in synthetic hsa-Let-7a-5p miRNA in 50 ng RNA extract with respect to the respective synthetic miRNA sample alone. ΔCt were calculated as Ctsynthetic—Ct^{endogenous + synthetic}. Data determined from S15 Fig.