Highlights
-
•
Right heart catheterization hemodynamics guide heart failure and pulmonary hypertension care, but the accuracy of the site compared to core laboratory interpretations remains uncertain.
-
•
Our findings suggest that site and core laboratory readings of mean pulmonary artery pressure and pulmonary capillary wedge pressure were largely concordant.
-
•
Across key hemodynamic thresholds, about 10% to 13% of interpretations were underestimated, which could influence clinical decision-making, though likely mitigated by the broader clinical context in real-world practice.
ABSTRACT
An accurate interpretation of hemodynamics measured by right heart catheterization (RHC) is crucial for guiding the management of heart failure and pulmonary hypertension. We evaluated the concordance and variability between site-reported and core laboratory-adjudicated interpretations of invasive hemodynamic measurements.
Background
Right heart catheterization (RHC) is crucial for informing diagnosis and treatment decisions in heart failure (HF) and pulmonary hypertension (PH). , Current guidelines encourage research into noninvasive remote hemodynamic monitoring, while recognizing invasive RHC as the gold standard. Because the accuracy of RHC interpretations has not been rigorously assessed, we evaluated the concordance and variability of invasive hemodynamic data interpretation between clinical sites and a core laboratory (core lab).
Methods
A descriptive analysis was performed using de-identified data (August 2023-October 2024) from CAPTURE-HF (Machine Learning Model’s Performance for Noninvasive Intracardiac Pressure Monitoring in HF) a prospective, multicenter validation study of a noninvasive pressure sensor in all-comer adults undergoing RHC, where sites performed initial interpretations that were subsequently adjudicated by the core lab. Noninvasive sensor signals collected for the parent study were not included in these analyses. The dataset provided by the sponsor (Acorai, Helsingborg, Sweden) included de-identified demographics, site-, and core lab-interpreted invasive RHC measurements of right atrial pressure, pulmonary capillary wedge pressure (PCWP), and pulmonary artery systolic, diastolic, and mean pressures (mPAP; calculated as 2/3 diastolic + 1/3 systolic).
Standard procedural recommendations provided by the sponsor included internal jugular or brachial venous access with transducers being leveled and zeroed at the mid-chest line with the catheter tip at the same height. Measurements were to be obtained during free breathing, as recommended by ISHLT guidelines and sites were advised to minimize artifacts, such as catheter ringing and dampened signals. PCWP was measured as the end-expiratory mean crossing the A-wave over 3 respiratory cycles without ectopy. For atrial fibrillation, averaging across respiratory cycles reduces variability from irregular R-R intervals.
Digitized hemodynamic waveform images from 20 U.S. sites were independently reviewed by 2 blinded core lab experts at the Duke Clinical Research Institute using standardized protocols and digital calipers (Iconico, Inc. Philadelphia, PA). While CAPTURE-HF obtained patient consent, the Saint Luke’s Institutional Review Board waived additional consent for this secondary analysis.
Analyses focused on PCWP and mPAP, given their clinical importance in assessing pulmonary congestion and PH. RHC measurements were summarized by median and interquartile range (IQR). Site and core lab interpretations were compared using Pearson correlation coefficients, Bland-Altman analyses, and empirical cumulative distribution function (ECDF) curves of site minus core lab differences. Due to non-normal distributions, median differences and nonparametric limits of agreement (LOA; 2.5th and 97.5th percentiles) were reported. Median regression lines were overlaid on Bland-Altman plots to assess systematic bias. Misclassification rates between sites and the core lab were assessed using clinically meaningful thresholds: PCWP ≥18 mmHg (pulmonary congestion), mPAP >20 mmHg (diagnosing PH), and PCWP >15 mmHg (postcapillary PH), for which under- and over-estimation would be clinically significant. Hierarchical linear models were fit to estimate the proportion of variability (R 2) of site and core lab differences attributable to sites, operators, and selected patient characteristics (age, sex, body mass index, and RHC value). Exploratory analyses evaluated the association between body mass index and site versus core lab pressure differences.
Results
Among 1560 patients (mean age 63.2 ± 14.3 years; body mass index 28.7 ± 6.7; 37% women), site-reported values had a median (IQR) of 13.0 (8.0-20.0) mmHg for PCWP and 24.3 (17.3-34.3) mmHg for mPAP. Core lab-adjudicated values were slightly higher at 14.0 (9.0-21.0) mmHg for PCWP and 27.0 (19.0-37.0) mmHg for mPAP.
Correlations between site and core lab measurements were strong ( r = 0.94 for PCWP; r = 0.96 for mPAP). Bland-Altman analysis showed a median difference of −1.0 mmHg for PCWP with 95% LOA −6.0 to 4.0 mmHg ( Figure 1 A), and −2.0 mmHg for mPAP with 95% LOA −9.8 to 3.7 mmHg ( Figure 1 C), indicating modest variability.
Bland-Altman and ECDF of site vs core lab hemodynamic reads of PCWP and mPAP.
Abbreviations: ECDF, empirical cumulative distribution function; PCWP, pulmonary capillary wedge pressure; mPAP, mean pulmonary artery pressure. Panels A and C show Bland-Altman plots comparing site-reported and core lab-adjudicated measurements for PCWP and mPAP, respectively. The x-axis represents the average of site and core lab values, and the y-axis represents the difference (site minus core lab). The solid blue line indicates the median regression; horizontal red dashed lines mark the 2.5th and 97.5th percentiles. Panels B and D show ECDF curves for PCWP and mPAP, respectively. The x-axis represents the difference between site and core lab measurements (mmHg), where negative values indicate lower site measurements (underestimation) and positive values indicate higher site measurements (overestimation). The y-axis represents the cumulative proportion of patients with a difference of that magnitude or less. Vertical red dashed lines highlight example thresholds discussed in the text.
ECDF curves of site minus core lab differences for PCWP were small, tightly distributed, and centered near zero, with fewer than 10% of site-reported values ≥3 mmHg lower and fewer than 5% ≥3 mmHg higher than core lab-adjudicated values ( Figure 1 B). In contrast, differences in mPAP were more variable, with approximately 30% of site-reported values being ≥3 mmHg lower, while fewer than 5% were ≥3mmHg higher than core lab values ( Figure 1 D).
Table 1 summarizes misclassification rates at clinically relevant thresholds for diagnosing pulmonary congestion and PH. Underestimation occurred in 12.9% of cases for PCWP ≥18 mmHg, 9.6% for mPAP >20 mmHg, and 13.7% for PCWP >15 mmHg. Overestimation was less frequent, occurring in 3.2% of cases for PCWP ≥18 mmHg, 4.7% for mPAP >20 mmHg, and 4.5% for PCWP >15 mmHg.
Table 1
Misclassification of site vs core lab hemodynamic interpretations at clinically important thresholds.
| Threshold | Underestimation | Overestimation | Total reads |
|---|---|---|---|
| PCWP ≥ 18 mmHg | 69/ 535 (12.9%) | 32/ 994 (3.2%) | 1,529 |
| mPAP > 20 mmHg | 103/ 1,078 (9.6%) | 22/ 469 (4.7%) | 1,547 |
| PCWP > 15 mmHg | 88/ 643 (13.7%) | 40/ 886 (4.5%) | 1,529 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
