Harnessing the Power of microRNAs as Prognostic Biomarkers in Acute Heart Failure

This article refers to ‘Serially measured circulating microRNAs and adverse clinical outcomes in patients with acute heart failure’ by N. van Boven et al., published in this issue on pages 89–96. The discovery of microRNA over 20 years ago coupled with their stability in plasma has triggered investigation of circulating microRNAs as candidate biomarkers and/or as therapeutic targets in various diseases including heart failure. A growing body of evidence supports microRNAs as ubiquitous and key players in cellular function with regulatory roles in cardiac hypertrophy and fibrosis and in the deranged neurohormonal regulation of the circulation observed in heart failure. MicroRNAs add a layer of complexity to the response to cardiac injury or overload and to our understanding of how changes in gene network expression and related signalling pathways are regulated in heart disease. The plethora of microRNAs reported also point to many potential therapeutic targets for the amelioration of cardiac injury. Relatively few reports have addressed microRNAs as diagnostic and/or prognostic indicators in acute heart failure. Thus far, seven published reports focus on the use of microRNAs in prognosis and/or diagnosis in acute heart failure (Table 1).1-7 Correlation study on miR-16-5p to C-reactive protein, miR-106a-5p to creatinine, miR-223-3p to growth differentiation factor 15, miR-652-3p to soluble ST2, miR-199a-3p to procalcitonin and galectin-3 and miR-18a-5p to procalcitonin. 16 protein biomarkers and 12 AHF-specific miRNAs predictive of worse clinical outcome at 180-day mortality Discovery: 10 AHF, 10 CHF, 10 healthy controls with 6-month follow-up Validation: 9 AHF, 10 CHF, 9 COPD and 17 controls Extended cohort: 100 AHF at admission, 24 h, 48 h and day 7 12 miRNAs decreased in AHF vs. CHF vs. COPD vs. controls, 7 miRNAs (let-7i-5p, miR-18a-5p, miR-18b-5p, miR-223-3p, miR-301a-3p, miR-423-5p miR-652-3p) predictive for 180-day mortality Discovery: 236 AHF and 58 non-AHF, 44 stable CHF Validation: 711 AHF qRT-PCR on 5 miRNAs, normalized using spike-in controls miR-1, -126, -423-5p lower levels in AHF 423-5p associated with poor long-term outcome (1 year) Discovery: 32 AHF, 15 COPD and 14 controls Validation: 44 AHF, 32 COPD, 59 other breathless and 15 controls Discovery: 12 AHF, 12 controls Validation: 30 AHF, 20 non-AHF and 30 controls miR-423-5p predictor of HF (multivariate logistic regression including age and sex) with AUC of 0.91 miR-18b*, -129-5p, -1254, -675, -622 increased in AHF vs. controls In the current issue of the Journal, van Boven et al.8 have assessed the prognostic performance of 12 serial measured circulating microRNAs for prediction of 1-year outcomes in the TRIUMPH acute heart failure cohort. Both baseline and serial measurements of miR-1306-5p have significant prognostic power independent of other predictors including N-terminal pro-B-type natriuretic peptide (NT-proBNP). The findings with respect to miR-1306-5p were first developed from trans-cardiac blood sampling and tissue studies in an instrumented swine model. Demonstration of a cardiac source for this microRNA in the pig model was followed up with serial measurements of miR-1306-5p and 11 other microRNAs (previously reported as cardiac-enriched and/or muscle-specific) in 475 participants in TRIUMPH. The selection of microRNAs was rationale but clearly leaves a large number of other microRNA candidates unexplored. Notably, the investigators report very high rates of invalid or undetectable results for both miR-1306-5p and other heart-enriched microRNAs indicating the exploratory nature of this report and the challenges remaining before any prognostic microRNA panel can be introduced into routine clinical use. Given the large RNA sequencing data set generated and the previously reported studies (Table 1), the authors selected a very limited array of microRNAs for their exploratory clinical screening. Further work could usefully extend the selection of microRNAs to include more of those previously reported in acute heart failure (Table 1). Van Boven et al.8 are first to use normalize their data according to levels of plasma miR-486-5p. The authors reported that miR-486-5p was highly abundant (constituting 92.5–97% of all sequencing reads) in plasma samples. Notably, to date, there is no consensus on optimal normalizing strategies or choice of endogenous control to be used in studies of circulating microRNAs. A search on ‘miR-486-5p’ suggested this microRNA is significantly associated with body mass index and other measurements of obesity, particularly in children,9 and it is also correlated with ventricular contractility in the setting of right ventricular dysfunction.10 More evidence is required to warrant the use of miR-486-5p as a normalizer or endogenous control. It remains a challenge to establish the gold standard for normalizing circulating microRNA levels to ensure accuracy and reproducibility. For both the association of miR-1306-5p with prognosis and the use of miR-486-5p for normalization, the current findings require validation in further independent acute heart failure cohorts. Investigation of new candidate prognostic biomarkers (singly or in combination) requires assessment of what they might add to our existing predictive tools. The B-type natriuretic peptides, BNP and NT-proBNP,11 are established in this role and other peptide candidates include mid-regional pro-atrial natriuretic peptide (MR-proANP)12 and mid-regional pro-adrenomedullin (MR-proADM),13 which are known to add prognostic value to BNP and/or NT-proBNP.14 Other circulating markers with acknowledged independent prognostic power in acute heart failure include endothelin-1, soluble ST2, highly-sensitive troponin, growth differentiation factor 15, high-sensitivity C-reactive protein, galectin-3, and cystatin-C. Van Boven et al.8 report single and serial measures of miR-1306-5p are independently associated with all cause-mortality and heart failure hospitalization. However, miR-1306-5p does not add to the area under the curve, net reallocation improvement (NRI) or integrated discrimination improvement (IDI) for prognosis in models which already incorporate NT-proBNP. This further confirms, despite the likely pathophysiological importance of microRNAs in heart failure, their measurement is not yet warranted for ‘prime time’ clinical application. Given the prognostic performance of miR-1306-5p, the function of this particular microRNA in the setting of acute heart failure is of potential importance. The absence of any mechanistic insights is one tantalising deficit in the current report. What mechanisms underlay the association of circulating miR-1306-5p with disease progression in swine undergoing aortic banding or with prognosis in the human TRIUMPH participants? An in silico search using two microRNA-target prediction databases (miRDB and TargetScan) yielded >200 potential mRNA targets for miR-1306-5p with 52 targets appearing in both databases. These include aquaporin 1, which functions as a transmembrane channel with a role in water homeostasis,15 and HEY2 regulating cardiac electrical activity16 with variants associated with Brugada syndrome and high risk of sudden cardiac death. Further in vitro and in vivo functional studies are required to elucidate the role of miR-1306-5p. MicroRNAs have great potential to be developed as prognostic biomarkers. Much progress has been made but we are still far from any routine clinical application in heart failure. This paper adds new knowledge about miR-1306-5p and the report reflects a necessary pioneering step highlighting the many challenges on the way to defining the role of circulating microRNA in the understanding and management of acute heart failure. Conflict of interest: none declared.