Heart failure with preserved ejection fraction (HFpEF) is characterized by signs and symptoms of heart failure, including peripheral oedema, dyspnea and exercise intolerance, in the absence of a reduced left ventricular ejection fraction (LVEF $\ge$50%) [1]. Current understanding of molecular mechanisms underlying HFpEF is largely based on the 2013 “Paulus and Tschope paradigm” [2] relating coexisting comorbidities to myocardial remodelling and dysfunction, through a systemic pro inflammatory state. Non-cardiac co-morbidities such as diabetes, obesity, hypertension, and chronic kidney disease are common in HFpEF and have the ability to induce systemic inflammation. During inflammation, microvascular endothelial cells produce reactive oxygen species (ROS), which limits nitric oxide (NO) bioavailability leading to oxidative stress and endothelial dysfunction.

Oxidative stress and inflammation are closely interconnected. Transcription factors that regulate the expression of pro inflammatory cytokines are activated under oxidative stress conditions and in turn, induce the generation of ROS, thus creating a vicious cycle of oxidation and inflammation [3]. Myeloperoxidase (MPO), a leukocyte-derived enzyme, functions as a link between oxidative stress and inflammation. During inflammation, MPO is released and uses H${}_2$O${}_2$ as a substrate to produce hypochlorous acid, a powerful pro-oxidant and pro inflammatory molecule.

Studies suggest that plasma MPO levels are elevated in patients with HF compared to controls and that increasing levels of MPO are associated with restrictive diastolic stage, right ventricular systolic dysfunction and tricuspid regurgitation in HFrEF [4]. Furthermore, MPO may be involved in the pathophysiology of atrial fibrillation through atrial accumulation of MPO and consequent increase in fibrosis [5]. These results imply that MPO may be important also for the development of HFpEF where diastolic dysfunction, atrial fibrillation and fibrosis are major components. Indeed, a recent study showed that HFpEF patients displayed higher plasma concentration of MPO compared to healthy controls [6]. Furthermore, since oxidative stress and microvascular endothelial dysfunction are suggested as fundamental parts of the pathophysiology and development of HFpEF, MPO inhibition appears as an interesting therapeutic approach and a clinical trial investigating MPO inhibitor “AZD4831” (ENDEAVOR NCT04986202 and NCT03611153) is currently ongoing. Heterogeneity among patients with HFpEF has been singled out to explain the difficulty to find treatments improving prognosis in this population. Hence, identifying characteristics associated with high levels of MPO could be interesting to target subgroups of patients most likely to benefit from treatment with MPO inhibitors.

In this context, the objective of our study was to reinforce data about MPO elevation in HFpEF, to assess the relation between MPO levels and clinical parameters including vascular function and to determine patient characteristics associated with high levels of MPO.

The findings of this study are as follows: patients with HFpEF have higher levels of MPO than controls, MPO levels in HFpEF are positively correlated with inflammation (CRP levels), diastolic dysfunction (E/e’) and congestion (NT-proBNP) and negatively with left ventricular ejection fraction. Patients with MPO levels above the median suffer more often from diabetes, are more often males but tend to show less vascular stiffness (lower AIx) than patients with MPO levels below the median.

Several studies have shown a strong correlation between MPO and cardiovascular disease (CVD) including acute coronary syndrome, atherosclerosis, hypertension, and stroke [4, 15]. Consistently, recent studies that target MPO in animal models of CVD have demonstrated favourable outcomes with regard to disease progression [16]. However, data in HFpEF are limited to the study by Hage and collegues [6]. Our study corroborates their finding that MPO is elevated in HFpEF patients compared to controls and demonstrates that this applies also when the control group is older (74 $\pm$ 6 years) and with a proportion of women comparable to the HFpEF group (65 and 72% respectively). MPO levels showed moderate diagnostic value for HFpEF, less powerful in that regard than NT-proBNP levels (ROC curves Supplementary Fig. 1).

MPO-mediated oxidative stress may be one of the mechanistic link between comorbidities, inflammation and endothelial dysfunction at the source of HFpEF [3]. Comorbidities, namely obesity, diabetes, and ageing generate inflammation [17, 18, 19, 20], during which MPO is released and uses H${}_2$O${}_2$ as a substrate to produce hypochlorous acid, a potent pro-oxidant and proinflammatory molecule. MPO levels in our study were indeed correlated with markers of inflammation (CRP and NLR) and signs of myocardial remodelling, namely NT-proBNP levels, decreasing LVEF (although within the normal range) and increasing E/e’. It is still to be determined whether MPO plays a causative role in the development of the disease or if it is merely a bystander of neutrophils activation. Indeed, recent studies directly incriminate activated neutrophils in aggravating diastolic dysfunction in mice subject to pressure overload [21], and in HFpEF patients [22, 23].

High MPO levels were associated with diabetic status. This is not surprising since diabetes is known to promote a systemic pro-inflammatory state [24, 25]. Furthermore, MPO was shown to be predictive of insulin resistance in a population of obese patients [26]. Since the proportion of patients with diabetes was higher in patients with MPO levels above the median, and since the difference in pathophysiology between diabetic and non-diabetic HFpEF patients is a topic of interest [27, 28], we further investigated this association with logistic regression adjusted for age and sex. Interestingly, the combination of male sex and diabetic status seem particularly associated with higher levels of MPO among patients with HFpEF. Indeed, all men suffering from diabetes had MPO levels above the median, while the proportion was limited to 40% in the other subgroups (Fig. 4). This finding is consistent with the sex-specific proteomic profile of patients with HFpEF in the PROMIS study [29], where they demonstrated that inflammation-related pathways predominated in men.

On the other hand, we found no association between vascular stiffness or endothelial function and MPO levels. Even more surprising, vascular stiffness seemed less important in the patients with higher MPO levels (lower AIx, lower Ea). The augmentation index (AIx) is calculated from pulse waveforms as the ratio of the difference between the early and late systolic peaks of the waveform relative to the early peak (Fig. 1) and represents the relative importance of the reflected wave [30]. Multiple small reflections travel back to the proximal aorta and merge into a “net” reflected wave whose magnitude and timing depend on vascular stiffness. In older subjects, systolic wave reflections mediate late systolic load, with an important impact on LV relaxation [31, 32]. The augmentation index is not simply a measure of arterial stiffness and wave reflection, but was also shown to be elevated in conditions of increased LV contractility and may reflect overall ventricular-vascular coupling [33]. In HFpEF, high AIx was associated with abnormal LV diastolic responses to exercise, particularly in women, suggesting that arterial stiffness may contribute to the pathophysiology of HFpEF more commonly in women than in men [34]. The finding that patients with MPO levels above median do not display more endothelial dysfunction, nor vascular stiffness might be an indication that the sequence: comorbidities, inflammation, oxidative stress, endothelial dysfunction, myocardial remodelling is not straightforward. Rather, different mechanisms are probably involved in the development of myocardial remodeling and impaired vascular function, while both condition can ultimately lead to HFpEF. Recent data from phenomapping point towards the same direction. Indeed, although studies identify slightly different clusters depending on available variables [35, 36, 37, 38, 39], two clusters seem to be commonly differentiated: one with older patients with stiff arteries, small highly contractile LVs and high rates of electrical remodelling (atrial fibrillation) and the other with high rates of metabolic comorbidities, mainly diabetes, marked LV remodelling and advanced diastolic dysfunction. Inflammation and oxidative stress may play a more prominent role in the latter, hence the elevation of MPO (Fig. 5). Accordingly, there were more men and more patients suffering from diabetes in the group of patients with MPO levels above the median and they displayed lower (although $\ge$50%) LVEF and higher E/e’. These two subgroups might reflect two distinct pathophysiological mechanisms underlying HFpEF.

Fig. 5.

Illustration of patients’ characteristics associated with levels of myeloperoxidase below or above the median. Patients with heart failure and preserved ejection fraction and myeloperoxidase above the median are more often men, suffer more often form diabetes, show subtle left ventricular dysfunction and pronounced diastolic dysfunction (high E/e’) while patients with myeloperoxidase below the median are more often women with elevated vascular stiffness and high left ventricular ejection fraction.

We acknowledge this single centre study has several limitations. Maybe the most important arising from the small sample size. Unfortunately, restrictions related to the COVID pandemic interrupted the recruitment for several months. Furthermore, due to limitations of the EndoPAT technique, we could not obtain RHI and AIx for all patients. Despite our best effort to include controls of similar age and sex, both groups are not accurately matched for these characteristics. However, our groups are more alike than the only other published study demonstrating higher MPO levels in HFpEF [6]. The presence of a control group of similar age and sex is important since there are no validated reference values of MPO. On the contrary, there is a wide range of concentrations reported in the literature [40, 41, 42], hence standardisation will be necessary before routine use of MPO measurements.

In the context of the development of treatment with MPO inhibitor “AZD4831” (NCT03611153) it is interesting to note that not all patients might respond homogeneously. The results of our study suggest that patients with metabolic comorbidities, particularly diabetes, subtle LV dysfunction and evident diastolic dysfunction might benefit more from treatment targeting MPO while patients with predominant arterial stiffness (mostly females) and hyper contractile LV might be less responsive. Hence, while this study should be considered exploratory and hypothesis generating, it adds relevant information to existing literature. Future studies should aim at exploring the sex specific interplay between vascular inflammation and stiffness in this population, with special interest in features of metabolic stress such as obesity and diabetes.

Myeloperoxidase levels are elevated in HFpEF compared to controls, reflecting leukocyte activation and oxidative stress. Patients with levels of MPO above the median are more often males and suffer more often from diabetes. MPO levels in HFpEF are positively correlated with diastolic dysfunction and congestion and negatively with left ventricular ejection fraction. The association between oxidative stress and vascular stiffness, on the other hand could not be demonstrated and deserves future attention.

Data available on reasonable request.

SL recruited patients and controls, analysed echocardiographic studies, interpreted data and wrote the first version of the manuscript. AG performed the ELISA assay for myeloperoxidase. NM made substantial contributions to acquisition of data. DV, AP and BLG participated in data analysis and extensively revised the manuscript. SH and CB made substantial contribution to study design, acquisition of funding and extensively revised the manuscript. ACP designed the study protocol, acquired funding, was involved in data acquisition and interpretation and revised the manuscript. All authors read and approved the final manuscript.

The local ethics committee approved the study (approval number 2012/23AVR/199), and all subjects gave written informed consent before study enrolment (Clinical trial NCT03197350). The investigation conforms to the principles outlined in Declaration of Helsinki.

The authors wish to thank Florence Sinnaeve for the blood punctures as well as all the enrolled subjects for agreeing to participate in the study. Fig. 5 was created with https://biorender.com/.

This work was funded by a grant of the Fondation Nationale de la Recherche Scientifique of the Belgian Government (FRSM CDR 23597851). Dr Pouleur is supported by the FNRS as Clinical Researcher (40000507). Dr Lejeune is supported by Fondation Damman and Fondation Saint Luc for her fellowship.

The authors declare no conflict of interest.