Skip to main content

Table 3 Key points for the design and evaluation of miRNA biomarker studies

From: Are microRNAs suitable biomarkers of immunity to tuberculosis?

Topic Major pitfall/mistake(s) Strategies
• Cohort definitions • Insufficient inclusion criteria • Exact definition of criteria for infection/disease
• Donor/patient characteristics neglected
• Consideration of therapy/concomitant diseases
• Focus on well defined study groups (e.g. children with tuberculosis/LTBI with a known index case)
• Small study group • Insufficient statistical power due to multiple testing in `global' miRNA analyses • The definition of study group sizes markedly depends on (i) the number of miRNA candidates analyzed, (ii) the variability of target miRNA expression, (iii) the frequency of miRNA expressing target cells, and (iv) the desired sensitivity of the approach
• High variability of miRNA candidate expression due to disease-independent regulatory mechanisms
• Include as many of the before mentioned parameters for study group size calculations
• Cooperate with statisticians
• Tissue heterogeneity • Differences in the proportions of miRNA expressing cellular subset confound analyses • Usage of purified populations—as homogenous as possible
• Characterization of heterogeneity (e.g. by flow cytometry) to deconfound results of heterogeneous tissues
• Statistical design and methods • Application of inappropriate methods • The definition of biomarkers requires discrimination
• Significantly different is not the same as discrimination • Discrimination tests (e.g. support vector machines and linear discriminance analysis) include training and test steps and study groups need to be defined accordingly
• Selection of miRNA targets • Small study groups but extensive array analyses • Focus on selected miRNAs targets for small study groups → hypothesis-driven approach
• `Housekeeping' miRNAs • No comparable internal standards • Apply a group of `housekeeping' miRNAs used in previous studies