Discussion
Our pilot study is the first to prospectively evaluate cardiac fibrosis biomarkers and dysregulated miRNA in patients with HF undergoing CRT to determine their potential to predict functional response. It is the first study, to our knowledge, to profile specific miRNAs that are known to be dysregulated in patients with HFrEF. Preimplant, there were no significant differences between expressions in the prespecified biomarkers. None of the predefined biomarkers demonstrated ability to predict functional response; however, CTx and LBBB morphology trended towards significance. However, levels of PINP and miR-122 following CRT implantation were shown to vary significantly between responders and non-responders. Expression over time in both groups was shown to alter for MMP-2, MMP-9 and hs-TnT. Furthermore, specific changes in biomarker expression were shown to be associated with changes in functional, neurohormonal and LV geometry parameters both in short-term and long-term follow-up. The study also showed higher expression of cardiac fibrosis biomarkers systemically and higher expression of miRNAs in the coronary sinus.
Cardiac fibrosis biomarkers are known to be strongly associated with poor HF outcomes.2 Alteration in ECM turnover is a key feature of cardiac fibrosis and strongly associated with development and progression of HFrEF.3 Over the last decade, specific ECM biomarkers have been associated with poor HF outcomes2 and to predict response to CRT.4 Alteration in collagen synthesis and deposition, demonstrated by PINP/PICP for type I and PIIINP for type III, have been observed to predict response in several observational studies for functional and echocardiographic criteria. The exact behaviour and significance of collagen turnover have not been consistently replicated,4 principally due to variation in response definitions. No biomarkers in our study predicted response; CTx trended towards predicting response but did not reach significance. No statistically significant differences in ECM biomarker expression were demonstrated between echocardiographic responders or those with/without MACE. A substudy of the CARE-HF trial1 examined ECM biomarkers (PINP, PIIINP, ICTP, MMP-1) as potential predictors of response (survival and LVEF >35% at 18 months); they showed none predicted response, in keeping with our observations.16 More recently, The Markers and Response to CRT (MARC) study tested multiple pre-specified variables including ECM biomarkers (PIIINP, ICTP, PICP, MMP-9, TIMP-1) and they also showed none predicted CRT response (defined as ↓>LVESVi at 6 months).17 Our study demonstrated that reduction in PIIINP expression at 6 months was associated with greater 6MWT distance from baseline, suggesting reversal of cardiac fibrosis was associated with improved functional status, in contrast to observations made by Garcia-Bolao et al.18 PIIINP was observed to be more sensitive to changes in cardiac modelling implying it may represent a more long-term remodelling pattern. However, in the long term, these markers do not correlate with LV geometry parameters, therefore the precise relationship between collagen turnover and response to CRT remains unclear. We found that PINP, PIIINP and MMP-2 expression was higher in peripheral compared with coronary sinus samples. These observations are in agreement with findings of Tolsana et al,19 who reported MMP-2 was more highly expressed systemically than in the coronary sinus. Multiple cell types have been shown to secrete pro-MMP-2 that are not exclusive to the heart.3 The implications are that HF modifies systemic expression of MMP-2 alongside that of the heart. Furthermore, greater expression systematically suggests potential modification of cardiac ECM in HF is from non-cardiac sources. GDF-15 is a marker of myocardial stress and its ability to predict poor HF outcomes are well described. GDF-15 has been demonstrated to be a robust predictor of mortality following CRT.5 However, GDF-15 did not predict functional response (survival/HF hospitalisations, ↓>1 NYHA class or ↑>25% 6 MWD at 1 year);5 our observations support these conclusions. GDF-15 did not predict response and we did not observe any difference in baseline expression in those with and without MACE.
There is maladaptation of complex cardiovascular biological systems in HF. This maladaptation involves dysregulation of specific miRNAs which regulate and control these systems.6 MiRNA dysregulation has been associated with development of adverse cardiac remodelling in HFrEF.6 Marfella et al
7 observed altered expression in miRNA profiles between responders and non-responders using a miRNA microarray. Melman et al
9 recently identified that miR-30d was overexpressed in responders and had ability to predict response (increase LVEF >10% at 6 months). Neither study replicated the results of the other; however, the studies differed in methodology. Marfella et al
7 and Melman et al
9both used small cohorts with different characteristics and different quantification methods (microarray and quantitative PCR, respectively). We selected six specific miRNAs that have been demonstrated to be dysregulated in HFrEF, including miR-30d. MiRNAs were individually profiled by quantitative PCR methods, which is gold standard for quantification.6 None of the miRNAs were observed to be predictors of functional response and no variation at baseline was observed for echocardiographic responders. However, miR-30d had statistically higher expression in patients with no MACE at 12 months. This observation has not previously been made and may imply increased LV wall stress is protective; however, this has not been tested in a prediction model against other variables. Over the observation period, miR-122 expression was statistically lower in functional responders. Recently, miR-122 has been shown to be expressed in the liver due to congestion which would support the observation that there is lower expression in responders after CRT.20 Higher expression of miR-30d and miR-133a was observed in the coronary sinus compared with peripheral circulation consistent with previous findings that demonstrated these miRNAs to be enriched.6 MiR-486 demonstrated higher expression in the coronary sinus than peripherally, potentially reflecting higher haemolysis of samples taken via the catheter.
Our cohort behaved as expected following CRT implantation.1 18 19 Overall, patients had significantly improved NYHA classification and 6MWT distances, alongside reduced Qualiy of Life (QoL) scores. Interestingly, the entire cohort showed a statistically significant improvement in LV geometry and reduction in hs-TnT over the observation period, without any difference between responders and non-responders. The pattern observed in overall improvement in LV geometry has been well described following CRT.1 However, given this pattern was not seen in functional responders only emphasises the known poor correlation between echocardiographic and clinical response criteria.21 Variation in different definitions used for CRT response remains a major limitation in this field.21
There are several limitations to our prospective study. The first and most important is that our study was underpowered and may have affected the study findings. We were unable to perform power calculations at the start given the lack of data in the field. We performed this single-centre pilot study to allow us to subsequently perform a power calculation which would inform a larger multicentre trial. Therefore, a larger, powered sample size may demonstrate different results. Transthoracic echocardiography was limited in several participants due to body habitus resulting in inability to perform paired standardised modified Simpsons biplane assessments on all (28 paired assessments performed). The NICE implant guidelines10 used changed during the study period reflecting a move away from mechanical dys-synchrony on echo towards QRS duration/morphology. The study inclusion changed to reflect real world circumstances and this may have altered cohort characteristics. Also, the number of MACE events was low over 12 months and results must be interpreted cautiously. Coronary sinus samples were taken on the second half of the cohort only. Given the small sample size caution must be applied when interpreting these results. It is possible there may be a delay in ECM biomarkers/mRNA levels following CRT and longer follow-up may cause changes undetected during our follow-up.