Biomarkers have been described as characteristics that can be objectively measured and evaluated as indicators of normal biological processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention. Biomarkers have been shown to have clinical utility for improving the diagnosis, treatment options and prognosis of patients. In respiratory medicine, molecular biomarkers have contributed to the comprehensive clinical management of diseases (e.g., circulating biomolecules and the expression of certain genes), and many biomarkers are either currently in use or under investigation. Some of the most important contributions of biomarkers to respiratory medicine are elevated levels of fibrinogen, C-reactive protein, and leukocyte count, which have been found to be associated with an increased risk of exacerbations in chronic obstructive pulmonary disease (COPD) patients.1 Similarly, assessing procalcitonin levels may help distinguish lower respiratory tract infections induced by bacteria from those that are not induced by bacteria, thereby preventing unnecessary antibiotic treatments.2 Sputum eosinophilia can be useful for determining asthma endotype or phenotype. Although a multitude of potential biomarkers for the personalized management of respiratory pathologies have been explored, not all have been found to be useful. Ideally, a biomarker must fulfill several requirements to be deemed useful. Mainly, it needs to be specific to the condition or disease being evaluated, available through minimally invasive means, quantitatively related to the extent of exposure, and cost-effective to assay. It is unusual for a single biomarker to meet all these requirements. In clinical practice, it is common to combine several biomarkers to diagnosis a disease, and give a prognosis, and predict response to a certain treatment. Therefore, in precision medicine, the best strategy would be to use a combination of biomarkers that can perform all these functions and are specific to the disease being evaluated. In summary, there is a need for further research into different biomarkers and their combinations to improve personalized medicine.

MicroRNAs (miRNAs) are noncoding RNA molecules with lengths ranging from 19 to 25 nucleotides that have been implicated in many biological processes. miRNA dysregulation is widely involved in the pathophysiological processes of human diseases and has been identified as a promising biomarker. These molecules function as posttranscriptional regulators. The function of miRNAs begins with their binding to the target mRNA and repressing its translation or marking it for degradation. The fact that their levels have been found to be dysregulated in several pathologies and even in response to treatments, as well as the fact that they are present in samples obtainable through liquid biopsies3 makes them very attractive candidates as potential biomarkers. Previous studies have explored the potential of miRNAs as biomarkers in different pathologies. Regarding lung cancer, miR-20a, miR-10b, miR-150 and miR-223 have been proposed as potential biomarkers for non-small cell lung cancer, and miR-205 has been proposed as a potential biomarker for squamous cell carcinoma.4 Another study evaluating miR-7a methylation levels in chronic obstructive pulmonary disease (COPD) patients found higher levels of methylation in COPD patients with the emphysema phenotype, who are generally at a higher risk of suffering lung cancer.5 Interestingly, altered levels of miR-320b, miR-22-3p and miR-423-5p have been reported in extracellular vesicles isolated from bronchoalveolar fluid lavage (BAL) of COPD patients compared to healthy nonsmokers. Moreover, miR-100-5p expression differs between healthy smokers and COPD patients.6 Extracellular vesicles isolated from BAL have also been used to perform similar studies in patients with lung sarcoidosis, with levels of miR-150 and miR-146a being upregulated in patients with parenchymal involvement (stage II) compared to patients with bilateral hilar lymphadenopathy only (stage I).7 In childhood asthma patients, miR-155-5p and miR-532-5p have been associated with lung function recovery in response to inhaled corticosteroids.8 The heterogeneity of asthma causes treatment outcomes to differ, thus increasing the need for biomarkers that can discern between different endotypes and predict and monitor treatment. A study explored miRNA expression in extracellular vesicles isolated from obese patients with obstructive sleep apnea (OSA) and obese hypoventilation syndrome, and found that vesicles isolated from these patients could induce endothelial dysfunction in vitro. This effect was no longer observed when extracellular vesicles were isolated from the same patients after receiving continuous positive airway pressure (CPAP) therapy. The same study reported changes in miRNA (miR-16-5p, miR-4459, miR-451a and miR-6510-5p) levels after treatment with CPAP.9 Another study showed the predictive capabilities of three miRNAs (miR-378a-3p, miR-486-5p, and miR-100-5p) to predict favorable blood pressure responses to CPAP therapy in patients with resistant hypertension and OSA.10 Since OSA patients are at increased risk of cardiovascular disease, miRNAs could help predict adequate response to CPAP therapy and identify OSA patients who could benefit from such treatment. Levels of miR-328 were found to be upregulated in a rat model of pulmonary fibrosis. The silencing of miR-328 was also found to alleviate the progression of pulmonary fibrosis. Interestingly, the same study found that miR-328 directly targets and inhibits FAM13A, a gene whose polymorphisms are known to be associated with prognosis in idiopathic pulmonary fibrosis.11 If these findings were replicated in humans, levels of rno-miR-328's ortholog could potentially be used to monitor disease progression and perhaps even treatment response. Of note, the existence of a miRNA-based signature that could improve risk stratification in patients with COVID-19 has been reported. This signature consists of 4 blood miRNAs (miR-16-5p, miR-192-5p, miR-323a-3p and miR-451a) that predict mortality in early-in-ICU patients with COVID-19.12 miRNA screening has also been combined with machine learning approaches for biomarker discovery. Using four different machine learning models in combination with the miRNA profiling of pulmonary hypertension patients, a study managed to screen out two miRNAs (miR-636 and miR-187-5p) able to diagnose the condition with high accuracy.13 Likewise, analysis through machine learning of miRNA content isolated from extracellular vesicles identified subjects at risk of accelerated lung function decline.14

Overall, research on miRNAs and their potential as biomarkers has gained traction over the past decade. The examples presented here are summarized in Table 1. Many studies have shown how miRNA levels vary in different pathologies or in response to treatment. Nonetheless, further research is needed to fully exploit their role as biomarkers. If the mechanisms through which miRNAs are involved in pathophysiological processes are unknown and the identified miRNA signatures in these studies are not validated in large enough and independent cohorts, it could lead to incorrect interpretations.

Combining miRNA profiling with innovative artificial intelligence-based tools will accelerate biomarker discovery considerably. Remarkably, artificial intelligence-based systems may have the greatest impact in the implementation of biomarkers in clinical prognosis and prediction.15