The predictive capabilities of a novel cardiovascular magnetic resonance derived marker of cardiopulmonary reserve on established prognostic surrogate markers in patients with pulmonary vascular disease: results of a longitudinal pilot study

Pulmonary arterial hypertension (PAH) is characterised by a pathological increase in the resistance of the pulmonary circulation secondary to arterial wall remodelling, vasoconstriction and in situ thrombosis (pulmonary vascular disease, PVD). Progressive obliteration of the pulmonary vascular bed eventually leads to right ventricular (RV) failure and premature death. Therapeutic advances have contributed to better long-term outcomes but disease progression remains difficult to predict and objectively monitor necessitating new methods.

We previously demonstrated proof-of-concept for a novel non-invasive method to assess ‘cardiopulmonary reserve’ as it pertains to PAH by measuring the average pulmonary arterial blood flow velocity (meanPAvel) at rest and during standardised intravenous (IV) adenosine infusion using phase-contrast cardiovascular magnetic resonance (CMR) [1]. This approach to ‘stressing’ the cardiopulmonary unit, whilst novel, was feasible, safe and simple, with the results confirming that meanPAvel at peak hyperemia was an excellent functional correlate for cardiopulmonary reserve across a range of clinical risk phenotypes.

As a marker of cardiopulmonary reserve and with the advantage of ameliorating the impact of unrelated systemic processes on variables measured at rest, we hypothesised that meanPAvel at peak hyperemia may provide prognostic information at initial assessment and during follow up. To evaluate this, we investigated the association of meanPAvel at rest and during peak hyperaemia with accepted RHC-derived, CMR-derived, biomarker and clinical prognostic surrogate markers in a cohort of patients with confirmed PAH or at high risk for incident PAH. Specifically, we hypothesised that relative to meanPAvel at rest, meanPAvel at peak hyperaemia at initial assessment would be more closely associated with comparator parameters at initial and follow-up assessments, and that changes in meanPAvel at peak hyperemia would correlate more closely with changes in these prognostic markers over time.

We prospectively recruited participants with known or suspected PAH and a clinical indication for a right heart catheter (RHC) through a single tertiary-referral centre over a 14-month period (2013–2014). Patients were ineligible if they were <18 years old; pregnant; had known or suspected PAH due to congenital heart disease with left-to-right shunt, or portopulmonary hypertension; or, had a contraindication to CMR or IV adenosine. Right heart catheterization for suspected PAH was deemed clinically indicated by pulmonary hypertension experts following a comprehensive review of clinical status and ancillary investigations, in accordance with guidelines [2]. All participants provided written informed consent and protocols were approved by the Local Institutional Research Ethics Committee.

At initial assessment, all participants underwent a RHC and CMR with IV adenosine within 48 h, with PAH confirmed by hemodynamic criteria (mean pulmonary arterial pressure, mPAP, >25 mmHg; mean pulmonary arterial wedge pressure, mPAWP, <15 mmHg; and pulmonary vascular resistance, PVR >3 Woods units, WU). Participants with mPAWP > 15 mmHg were excluded and did not undergo vasoreactive testing. Patients with PAH were invited to undergo repeat RHC and CMR with vasoreactivity testing and participants with suspected but excluded PAH (high-risk for incident PAH) invited to undergo repeat CMR with vasoreative testing approximately 6 months after their initial study. Between assessments, patients were administered guideline-appropriate management.

Right heart catheterization was performed via the right femoral vein with a 7 F Swan-Ganz thermodilution pulmonary artery catheter (Edwards Lifesciences, California, USA). Systemic arterial pressures were monitored by arm cuff. Complete resting hemodynamic assessment was performed in accordance with current guidelines. Intravenous (IV) adenosine (Adenoscan, 3 mg/mL, Astellas Pharma, Tokyo, Japan) was infused via peripheral vein at three increasing doses: 70-, 140-, and 210mcg/kg/min, and repeat hemodynamic assessment (mPAP/mPAWP/CI/PVR) conducted after a minimum of two minutes’ infusion at each dose.

CMR was performed with a 1.5 T magnet (Magnetom, Siemens, Erlangen, Germany) with images obtained at end-expiration. Volumetric and functional analyses were performed using steady-state free procession (SSFP) sequences of the atria and ventricles. Flow imaging through the main pulmonary artery was performed 1.5-2 cm above the pulmonary valve using a velocity-encoded gradient echo sequence with an upper velocity limit of 150 cm/s, temporal resolution of 39 ms and spatial resolution of 1.8x1.8x6mm (Fig. 1). Adenosine was infused using the same protocol as during RHC, with flow imaging through the pulmonary trunk repeated after a minimum of 2 min at each adenosine dose. Analyses were performed offline using specialized software (CMR42, Circle Cardiovascular Imaging Inc., Calgary, Canada). Ventricular endocardial and epicardial contours were manually outlined at end-diastole and end-systole to permit calculation of ventricular volumes and myocardial mass by Simpson’s method. Blood flow through the pulmonary trunk was measured by outlining the endovascular border at all 20 reconstructed cardiac phases permitting calculation of mean pulmonary arterial blood flow velocity (meanPAvel), elastic modulus (pulse pressure x minA/(maxA-minA)) and a volumetric estimation of ventriculoarterial (VA) coupling (RVESV/RV stroke volume) [3]. All participants underwent 6-min walk tests and had blood samples taken for NT-proBNP quantification, which was centrifuged within 1 h for 10 min at 2800 rpm, and stored at −80 °C until analysis using an ELISA kit (USCN Life Science, Texas, USA).

Thirty-one participants were recruited (PAH n = 19; suspected but excluded PAH: ‘High Risk’ n = 12) of which 5 were ineligible (mPAWP > 15 mmHg n = 4; claustrophobia n = 1). 20 underwent repeat investigations (PAH n = 14; High Risk n = 6) after a median period of 8 months and 21 days (loss to follow-up in PAH group: death from RV failure n = 1; geographical relocation n = 1; declined n = 1, and High Risk group: geographical relocation n = 2; intercurrent illness n = 1) and were included in the present study. Ten of the 14 PAH participants had both repeat CMR and RHC and 4 had CMR only (declined repeat RHC n = 4). Demographic and clinical characteristics of participants are presented in Table 1, with suspected but excluded PAH participants considered high risk for incident PAH on the basis of the high prevalence of connective tissue disease, unaccounted for exertional breathlessness with disproportionately reduced lung diffusing capacity, elevated NTpBNP levels in the absence of left ventricular myocardial disease or renal dysfunction, elevated estimated pulmonary arterial systolic pressure by transthoracic echocardiography >40 mmHg (n = 6) and ‘borderline’ abnormal resting hemodynamics (n = 2 with rest mPAP of 21-24 mmHg) [5].

Mean absolute values of comparator, investigational and other relevant parameters, and their relative changes over time, are summarised in Table 2.

Blood flow velocity through the main pulmonary artery is thought to decline in PAH due to impeded passage of cardiac output through remodelled microcirculation, dilation of proximal pulmonary vessels and eventually, impaired RV function [7]. Not surprisingly therefore, mean pulmonary arterial blood flow velocity (meanPAvel) measured at rest has been shown to correlate with measures of RV function, arterial load and exercise capacity, which are all important determinants of prognosis in PAH [7‐11]. This study adds to our knowledge by demonstrating correlation between our non-invasive measure of ‘cardiopulmonary’ reserve (meanPAvel at peak hyperemia, [1]) and validated invasively and non-invasively derived surrogate prognostic markers at baseline and six month follow-up. This marker was superior to meanPAvel when measured at rest. This data would support further studies being undertaken with hard clinical endpoints, to establish if this novel marker would provide incremental prognostic information in patients with PAH or at high risk for incident PAH.

PAH is principally defined by the plexogenic pulmonary vasculopathy that underlies its genesis, increasing RV afterload and stress. These two compartments (the RV and pulmonary vasculature) are integrally related and not surprisingly therefore, measures of RV afterload (static e.g. PVR, and oscillatory e.g. pulmonary arterial stiffness, components), stress (e.g. NTpBNP), and functional adaptation (e.g. RVEF, RVESVI) all provide independent prognostic information. In isolation however, each parameter tends to provide a limited picture of the state of the ‘cardiopulmonary’ unit as a whole. In our study, meanPAvel at peak hyperemia correlated significantly with measures of RV afterload (e.g. PVRI and elastic modulus), RV functional adaptation (e.g. RVEF/RVESVI) and VA coupling (defined as the ratio of maximal systolic RV elastance: effective arterial elastance), suggesting a capacity to simultaneously interrogate the pulmonary vasculature, the RV, and their interaction in a simple, non-invasive fashion. The capacity to assess the RV and pulmonary vasculature as a collective unit may translate to improved ability to discriminate changes in ventricular function, arterial load, or both when compared to, for example, RVEF, a serial CMR measure that conveys consistent prognostic information in PAH patients [3, 12, 13].

‘Traditional’ prognostic parameters are generally measured at rest and are subject to the influence of unrelated systemic factors such as anxiety, hypertension and sedation which may be compounded by serial assessment. Moreover, they lack the capacity to assess physiological reserve, that is, they provide a static measure of a dynamic disease process. Our protocol, like other physiological stress tests, negates the influences of unrelated systemic factors whilst permitting assessment of reserve as it pertains to PAH, that is, the degree of functional pulmonary circulation recruitable by endothelial independent mechanisms with IV adenosine (‘vascular reserve’) and the capacity of the RV to increase blood flow in response to downstream circulatory recruitment (‘RV reserve’: chronotropic/inotropic). We postulate that the generally closer association found between meanPAvel at peak hyperemia with validated prognostic parameters, compared with meanPAvel measured at rest, was due to the capacity to assess the cardiopulmonary unit as whole and its’ reserve while ameliorating the impact of unrelated systemic processes.

Acknowledging the above, not all validated prognostic comparators correlated significantly with meanPAvel at peak hyperemia. Most notably, NTpBNP levels measured at initial assessment did not correlate with meanPAvel measured at the same assessment or at follow-up, although the relative change in NTpBNP correlated with the change in meanPAVel at peak hyperemia over time. Previous studies have demonstrated association between baseline NTpBNP levels, resting hemodynamics, lung function, peak VO2 and prognosis, CMR-derived measures of RV function, and association between change in NTpBNP over time and survival [14‐19]. The discordant findings of strong correlation between meanPAVel at peak hyperemia, resting hemodynamics and markers of RV functional adaptation (RVEF, RVESVI) but no correlation with NTpBNP levels in absolute terms were unexpected and difficult to explain biologically: they may therefore reflect a statistical anomaly. A lack of correlation between change in meanPAvel and change in PVRI over time was likely due to the small sample size undergoing repeat RHC (N = 10) or, as earlier discussed, the impact of unrelated systemic factors on resting measures that may compound with serial assessment. No correlation between change in meanPAvel at peak hyperemia and change in 6MWD over time may reflect the inherent limitations of serial 6MWDs and is in keeping with registry data showing a decline in 6MWD is prognostically relevant whereas an increase is not [6]. Closer association between these variables in the subset of patients that experienced a decline in 6MWD (n = 9: R = 0.59, P = 0.22) is in keeping with our other findings, acknowledging that statistical significance was not met, likely due to a small sample size.

Mean pulmonary arterial blood flow velocity measured using CMR during non-invasive standardised pulmonary vasoreactive testing correlated moderate-strongly with a range of CMR-derived, RHC-derived, biomarker and clinical prognostic variables at initial assessment and at follow up. Change in meanPAvel at peak hyperemia on repeat vasoreactivity testing correlated with change in these variables over time, although correlation was generally weaker. Compared with meanPAvel measured at rest, correlation was generally stronger with meanPAvel measured at peak hyperemia. The novel parameter of meanPAvel at peak hyperemia may provide prognostic information by simultaneously interrogating the pulmonary vasculature and RV functional capacity whilst ameliorating the impact of unrelated systemic processes, although larger trials are required to confirm these findings.