Non-invasive assessment of stroke work using cardiac magnetic resonance imaging

Gareth Matthews, Ciaran Grafton-Clarke, Andy J. Swift, Alisdair Ryding, Hosam Assadi, Pankaj Garg, Rui Li, Zia Mehmood, Chris Sawh, Rebecca Gosling, Samer Alabed

Research output: Contribution to journalAbstractpeer-review


Introduction Although widely adopted in clinical practice, the left ventricular ejection fraction (LVEF) has significant limitations in heart failure assessment1. Gold standard physiological measures of heart function utilise pressure-volume (PV) loops, particularly the stroke work (SW), represented as the area bounded by the loop. Invasive assessment of PV-loops precludes its wide-spread clinical use. Cardiac magnetic resonance (CMR) imaging allows the accurate assessment of volumetrics and recent studies have found it useful in estimating left ventricular filling pressure2. We hypothesise that non-invasive parameters can be used to accurately estimate SW.

Methods In the Wellcome Trust PREFER-CMR study (NCT05114785) invasive PV-loops were collected using impedance catheters (7-Fr PV-Loop Catheter CA-71103-PL; CD Leycom, NL) from 8 patients who also underwent CMR assessment within 4 hours, which were then analysed using custom algorithms in MATLAB. A non-invasive, physiological method of SW assessment was developed using CMR parameters and the peripheral blood pressure. This method was subsequently applied to a large retrospective cohort of 835 heart failure and pulmonary hypertension patients who had undergone both CMR and invasive pressure assessment at the Sheffield Pulmonary Vascular Disease Unit2. Correlations were assessed using the Pearson Correlation Coefficient. Variables were normalised prior to backward conditional multivariate Cox regression. All work was received ethical approval (REC: 21/NE/0149, 16/YH/0352).

Results CMR SW correlated with that from invasive PV-loops (typical loop shown in Figure 1) in the small discovery impedance catheter cohort (r=0.861, P=0.006) and no significant difference demonstrable (Mean invasive SW 1.298±0.13J vs. CMR SW 1.228±0.14J, paired t-test p=0.350). Non-invasive CMR SW significantly correlated with hybrid invasive/CMR-derived PV-loop SW (r=0.995, p<0.001). Univariate regression analysis demonstrated significant risk of all cause death associated with reduced CMR SW, increased CMR-derived pulmonary capillary wedge pressure (PCWP) and reduced LVEF, with CMR SW having the greatest effect size (Table 1). On multivariate regression, age, male sex, CMR SW and CMR PCWP remained independently predictive of overall death whilst LVEF did not (Table 1). For 1-year mortality, only CMR SW age and male sex remained significant, with CMR-derived PCWP and LVEF excluded (Table 2). ROC curve analysis gave an area under the curve of 0.634 for CMR SW in isolation and 0.767 for the multivariate model (Figure 2A) for overall mortality. Using the multivariate regression, a risk score based on age, sex, CMR SW and CMR PCWP with cut-off value of >1.45 is shown in Figure 2B demonstrating significant difference in survival.

Conclusions A combination of simple blood pressure assessment and CMR-derived volumetrics were able to accurately estimate the cardiac SW which was predictive of both 1-year and overall mortality. Such a method is amenable to high throughput and artificial intelligence assisted clinical application for heart failure management.
Original languageEnglish
Pages (from-to)A200-A201
Number of pages2
Issue numberSuppl 3
Publication statusPublished - 2 Jun 2023

Cite this