Non-invasive methods for estimating mPAP in COPD using cardiovascular magnetic resonance imaging

Pulmonary hypertension is a predictor of death and hospitalisation [1‐5] in patients with chronic obstructive pulmonary disease (COPD). Patients with pulmonary hypertension in COPD, as defined by a mean pulmonary artery pressure (mPAP) of ≥25 mmHg have a 5-year survival rate of 36% [6]. A number of studies in patients with severe COPD have shown that mild pulmonary hypertension is common: a recent study in patients with severe COPD referred for lung volume reduction surgery demonstrating pulmonary hypertension at right heart catheter in 50% [7, 8]. More recently, with the advent of therapies for other forms of pulmonary hypertension there has been increasing interest in the subset of lung disease patients with severe (previously called “out of proportion”) pulmonary hypertension, defined as an mPAP ≥35 mmHg or mPAP ≥35 mmHg with cardiac index ≤2.0 [9], where a cardiovascular limitation to exercise, rather than respiratory limitation exists [10]. This raises the possibility that pulmonary vascular therapies may improve symptoms and outcome in this patient group [11].

The gold standard for diagnosis of pulmonary hypertension is right heart catheter (RHC), however, this is an invasive test [12]. As such, patients are screened for pulmonary hypertension with echocardiography, but unfortunately, this is less accurate in COPD where pulmonary artery pressure, when measured, is often overestimated [13, 14]. A non-invasive method for estimating mPAP in COPD patients would, therefore, be useful to help diagnose PH, in prognostication and for possible assessment of treatment response or follow-up in clinical trials.

Several predictive cardiac magnetic resonance (CMR) imaging models have been proposed for estimation of pulmonary artery pressure [15‐17]. The predictive value of these MR derived imaging models in a population of patients with suspected pulmonary hypertension in COPD remains unknown. The aim of this study was to assess the diagnostic accuracy and the prognostic value of these published models of non-invasive mPAP prediction using cardiovascular MRI.

All consecutive patients who underwent MRI at a pulmonary hypertension referral centre [18] from April 2012 to October 2015 with suspected pulmonary hypertension were assessed for inclusion. Inclusion criteria were a formal diagnosis of COPD (according to standard criteria), as per the ASPIRE (Assessing the Spectrum of Pulmonary hypertension Identified at a REferral centre) registry [18]. Patients were assessed for either obstructive spirometry (defined as FEV1/FVC ratio of ≤0.70) or qualitative CT evidence of emphysema, as per standard radiological practice [19]. Any patients without RHC and MRI within 90 days were excluded. Ethical approval was granted from a local ethics committee for this retrospective study, written consent was waived (ref c06/Q2308/8).

A total of 1864 patients were referred to the pulmonary hypertension centre during the period studied and underwent right heart catheterisation: of these 145 had a documented diagnosis of COPD from a respiratory specialist. One hundred and two had MRI and RHC within 90 days so were included in the study. There were 87 patients with pulmonary hypertension (69 of these had severe pulmonary hypertension) and 15 patients without pulmonary hypertension (please see Fig. 1 for the patient diagnostic pathway). Figure 2 shows the flow of patients within the diagnostic pathway. There were 24 with GOLD severity 1, 40 with GOLD severity 2, 18 with GOLD severity 3 and seven with GOLD severity 4; 13 patients did not have spirometry results available for analysis. Of the 102 cases of COPD included in the study, 75 had spirometric evidence of airflow limitation (FEV1/FVC ratio <0.7), and the remaining 17 had evidence of emphysema on CT. Table 1 shows the patient demographics and clinical characteristics for the whole group and split into no PH (mPAP <25 mmHg), PH (mPAP 25 to ≤35 mmHg) and severe PH (mPAP ≥35 mmHg).

This study demonstrates that CMR models can be used to assess for the presence of pulmonary hypertension in COPD with good diagnostic accuracy, whilst also providing prognostic information similar to right heart catheterisation. This data supports the potential use of CMR in the non-invasive assessment of suspected pulmonary hypertension in patients with COPD. We have shown that single measurements from cardiac MRI (most notably septal angle) correlate strongly with right heart catheter measured mPAP. A model which includes measurements from the right ventricle, the diagnostic accuracy is slightly reduced, but inclusion of basic measurements of pulmonary arterial structure and function increases the accuracy of predicted mPAP and, therefore, the presence of pulmonary hypertension. Therefore, we recommend the use of CMR-PA/RV model in the diagnosis of pulmonary hypertension in patients with COPD. In the presence of COPD, right ventricular mass index and PA relative area change are the strongest predictors of outcome, but further work to develop prognostic thresholds is required.

Studies performed primarily in patients with pulmonary arterial hypertension have shown that structural and functional indices of the right ventricle (RV end diastolic volume index, RV end-systolic volume index, RV ejection fraction, RV mass, interventricular septal angle) and pulmonary artery (PA area, PA relative area change and black blood score) correlate with RHC measured mPAP. Inter- and intra-observer reproducibility data for these cardiac MRI metrics for DC and AJS have been previously published. RV end-diastolic and systolic measurements had excellent inter-observer reproducibility (ICC 0.959 and 0.991) and pulmonary artery relative area change had high inter-observer reproducibility (ICC 0.891). Right ventricular ejection fraction also had high inter-observer reproducibility (ICC 0.957) [22]. We and others, have shown that we can further develop predictive equations using MR metrics to estimate measurements made at cardiac catheterisation. However, to date there is very little data on the use of MR imaging as a diagnostic and prognostic tool in patients with COPD and suspected pulmonary hypertension. In this study we have shown that these structural and functional indices of the right ventricle and pulmonary artery correlate with right heart catheter measured mean pulmonary arterial pressure: with pulmonary artery relative area change, ventricular mass index and interventricular septal angle having the strongest correlations. We have shown that the use of CMR-PA/RV and CMR-RV models, improve the prediction of mPAP and also the diagnostic accuracy of quantitative CMR measures.

The alpha model showed good correlation with mPAP, but had a low diagnostic accuracy: this model suffered from a diagnostic threshold that was too low for this cohort, significantly reducing its specificity. Changing the threshold of the alpha index to 16 increased the diagnostic accuracy of this model to 77% sensitivity and 80% specificity. This inaccuracy may be related to the measurement of right ventricular ejection fraction, particularly the reproducibility between different centres. In the original paper, it was not described how the right ventricular endocardial contours were measured, specifically, whether the papillary muscles and RV trabeculations were included, which may potentially add a bias into the calculation.

Pulmonary hypertension, confirmed by right heart catheter, has been recognised for many years as a marker of disease severity in COPD. In this study we have shown that cardiac MRI measured right ventricular mass and therefore ventricular mass index, along with pulmonary artery relative area change are significant predictors of mortality on univariate analysis. However, by using CMR derived parametric models, particularly the CMR-RV model which combine these CMRI derived measures, we can improve the prediction of outcome in COPD patients on multivariate analysis. These models may be further improved with novel imaging biomarkers in pulmonary hypertension assessment. One such marker is the 4D flow assessment of the life of vortices within the pulmonary artery, which has been shown to have a very high correlation with mPAP (r² 0.95) [23].

A potential source of error arises from the use of MR models, which were developed for use in a cohort of all causes of pulmonary hypertension and not validated in specific sub-groups, such as PH-COPD in this case. We feel that it is likely that the findings of a raised interventricular septal angle and ventricular mass index are likely to occur most subgroups of pulmonary hypertension (with the exception of left heart disease) as they are related to the pressure differential between the left and right ventricles. The measures of pulmonary arterial structure and function (size and relative area change) are likely to be transferable across subgroups, as they are markers of pulmonary vascular compliance and remodelling. We, therefore, feel that it reasonable to use these models in specific sub-groups of pulmonary hypertension, although validation, such as in this paper, would be useful. The models that are used in this paper use parameters that are stated with a degree of precision (for example an offset of 21.806 for CMR-PA/RV), likely more than is required for this purpose. We have maintained the equations in the published form to reduce any bias in the calculations, but it is likely that fewer decimal places could be used for these parameters for the prediction of outcome and the presence of pulmonary hypertension in COPD.

Currently, RHC is the gold standard test for the assessment of pulmonary hypertension and is used to assess prognosis in suspected COPD-PH cases. The ability to use non-invasive CMR models in this patient group may avoid the need for RHC, as we have shown CMR to be as good in the assessment of prognosis. The strong correlation of CMR imaging models with RHC mPAP suggest it may also have a role in follow-up, and also in assessment of treatment response in possible future trials assessing vasodilator response in COPD-PH. The role of cardiac MRI in COPD is further strengthened as this is a cohort of patients in which echocardiography is challenging. In the presence of COPD echocardiography has a relatively high rate of non-diagnostic quality studies. It has a good negative predictive value, but a poorer positive predictive value [14]. In patients with a normal right ventricle and predicted systolic pulmonary artery pressure (estimated from the tricuspid regurgitant jet velocity using the Bernoulli equation) pulmonary hypertension is highly unlikely. The main role for cardiac MRI in patients with COPD, therefore, probably lies in the group of patients with high estimated systolic pulmonary artery pressures or non-diagnostic scans. This may be at diagnosis, as studied in this paper, or potentially at follow-up.

The non-invasive estimation of mPAP has multiple other potential uses, beyond prognosis in COPD, although these have not been addressed in this paper. In advanced COPD patients considered for lung volume reduction surgery (LVRS), severe pulmonary hypertension is considered a contraindication [24], so the non-invasive assessment of mPAP on MR would be a useful tool in patient selection. Furthermore, left or right ventricular dysfunction is considered a relative contraindication to LVRS [24], and can be assessed at the same CMR sitting. In COPD patients with pulmonary hypertension there is evidence that outcome may be improved if treated with lung transplantation over LVRS [25], assessment of mPAP could be performed in the same sitting as a ventilation study in the preoperative assessment of LVRS or transplant candidates [26, 27], allowing for the assessment of the best surgical option and the best target in the lung.

This study is limited by its single centre, retrospective design, at a tertiary pulmonary hypertension referral centre where the severity of pulmonary hypertension is more severe. The results, therefore, represent a biased population, as the patients have been screened by echocardiography and excessive symptoms for the presence of pulmonary hypertension before referral (the vast majority of patients with COPD are managed in primary care). This has resulted in a smaller proportion of non-pulmonary hypertension patients than would be present in a general population of COPD; however, in the absence of a prospective trial, this is difficult to overcome. Furthermore, the assessment of the presence of COPD in the population relied upon the accurate documentation of a clinical diagnosis of COPD. This likely resulted in a lower proportion of patients being identified, but is likely to be more robust than using either spirometry or CT defined emphysema alone. Despite this, we feel that these tools are valid for the assessment of patients who are referred to a pulmonary hypertension centre regarding the presence of pulmonary hypertension and prognosis, as this is the population that was studied. It may be that more specific diagnostic thresholds would be suitable for use in the general COPD population.

The use of the CMR-PA/RV model, derived from cardiovascular magnetic resonance imaging in the assessment of COPD allows for accurate, non-invasive estimation of pulmonary artery pressure and the presence of pulmonary hypertension. Right ventricular mass index and pulmonary artery relative area change are most predictive of outcome.