Right Atrial Reservoir Strain by CMR Feature-Tracking Mirrors Pulmonary Vascular Load and Improves 6-Month Risk Prediction Beyond RVEF and NT-proBNP: An RHC-Anchored Cohort

Background: Right-atrial (RA) dysfunction may precede right-ventricular failure in pulmonary hypertension (PH). CMR feature-tracking (FT) can derive RA reservoir strain (RAε_res) from routine cine images. We assessed whether RAε_res reflects invasive pulmonary vascular load and adds prognostic value beyond standard metrics.

Methods: Single-center retrospective cohort of adults evaluated for suspected PH who underwent CMR and right-heart catheterization (RHC) within 30 days. RAε_res was measured on a 4-chamber cine using vendor-agnostic FT. Primary correlations: RAε_res vs pulmonary vascular resistance (PVR) and mean pulmonary artery pressure (mPAP). The primary outcome was a 6-month composite of HF hospitalization, escalation of PH therapy, or all-cause mortality. Multivariable Cox models adjusted for age, sex, etiology, NT-proBNP, RVEDVi, and RVEF. Reproducibility was assessed in 30 randomly selected scans.

Results: N=128 (56±14 years; 61% female). Hemodynamic profiles: pre-capillary PH 48%, post-capillary PH 39%, no-PH 13%. Median CMR-to-RHC interval: 21±9 days.
RAε_res (mean 26±9%) correlated with PVR (r=−0.68, p< 0.001) and mPAP (r=−0.62, p< 0.001). Over 6 months, 27/128 (21%) met the composite outcome. In multivariable analysis, lower RAε_res independently predicted events (per 5% decrease: HR 1.38, 95%CI 1.16–1.64, p=0.001). Adding RAε_res to a clinical model (age, sex, NT-proBNP, RVEDVi, RVEF) improved discrimination (C-statistic 0.72→0.81, Δ=0.09, p=0.010) and reclassified risk (continuous NRI=0.19, p=0.03). An RAε_res threshold ≤20% identified higher-risk patients (sensitivity 78%, specificity 72%). Inter-observer reproducibility was excellent (ICC 0.94).

Conclusion: CMR-derived RA reservoir strain closely mirrors invasive pulmonary vascular load and provides incremental, reproducible 6-month risk prediction beyond RVEF and NT-proBNP. Because RAε_res is obtained from routine cines without contrast or stress, it is a practical marker for PH phenotyping and short-term risk stratification.

Table 1. Hemodynamic and Clinical Profiles by RA Reservoir Strain Tertiles (Nf128).

Variable	Low RAεres ≤20% (n=41)	Intermediate 21–30% (n=45)	High >30% (n=42)	p-value
PVR (WU)	6.4 [4.9–8.2]	4.1 [3.1–5.0]	2.8 [2.1–3.6]	< 0.001
mPAP (mmHg)	41 ± 9	34 ± 7	26 ± 6	< 0.001
RVEF (%)	42 ± 8	48 ± 7	53 ± 6	< 0.001
NT-proBNP (pg/mL)	1850 [980–3220]	980 [520–1760]	520 [260–980]	< 0.001
6-mo Events, n (%)	14 (34%)	9 (20%)	4 (10%)	0.008

Table 2: Multivariable Predictors of 6-Month Composite Outcome (Nf128).

Variable (per unit change)	Adjusted HR (95% CI)	p-value
RAεres (per 5% ↓)	1.38 (1.16–1.64)	0.001
RVEF (per 5% ↓)	1.18 (1.01–1.39)	0.041
NT-proBNP (per log ↑)	1.31 (1.02–1.69)	0.033

Figure 1. RA Reservoir Strain Reflects Pulmonary Vascular Load and Predicts 6-Month Outcomes