Evaluation of Intracardiac Pressures Using Subharmonic-aided Pressure Estimation with Sonazoid Microbubbles

Abstract

Purpose

To investigate if the right ventricular (RV) systolic and left ventricular (LV) diastolic pressures can be obtained noninvasively using the subharmonic-aided pressure estimation (SHAPE) technique with Sonazoid microbubbles.

Materials and Methods

Individuals scheduled for a left and/or right heart catheterization were prospectively enrolled in this institutional review board–approved clinical trial from 2017 to 2020. A standard-of-care catheterization procedure was performed by advancing fluid-filled pressure catheters into the LV and aorta (n = 25) or RV (n = 22), and solid-state high-fidelity pressure catheters into the LV and aorta in a subset of participants (n = 18). Study participants received an infusion of Sonazoid microbubbles (GE HealthCare), and SHAPE data were acquired using a validated interface developed on a SonixTablet (BK Medical) US scanner, synchronously with the pressure catheter data. A conversion factor, derived using cuff-based pressure measurements with a SphygmoCor XCEL PWA (ATCOR) and subharmonic signal from the aorta, was used to convert the subharmonic signal into pressure values. Errors between the pressure measurements obtained using the SHAPE technique and pressure catheter were compared.

Results

The mean errors in pressure measurements obtained with the SHAPE technique relative to those of the fluid-filled pressure catheter were 1.6 mm Hg ± 1.5 [SD] (P = .85), 8.4 mm Hg ± 6.2 (P = .04), and 7.4 mm Hg ± 5.7 (P = .09) for RV systolic, LV minimum diastolic, and LV end-diastolic pressures, respectively. Relative to the measurements with the solid-state high-fidelity pressure catheter, the mean errors in LV minimum diastolic and LV end-diastolic pressures were 7.2 mm Hg ± 4.5 and 6.8 mm Hg ± 3.3 (P ≥ .44), respectively.

Conclusion

These results indicate that SHAPE with Sonazoid may have the potential to provide clinically relevant RV systolic and LV diastolic pressures.

Keywords: Ultrasound-Contrast, Cardiac, Aorta, Left Ventricle, Right Ventricle

ClinicalTrials.gov registration no.: NCT03245255

© RSNA, 2024

Summary

Using the subharmonic-aided pressure estimation technique with Sonazoid microbubbles, the right ventricular systolic and left ventricular diastolic pressures were comparable with measurements obtained using the cardiac catheterization technique.

Key Points

• ■ Subharmonic-aided pressure estimation (SHAPE) with Sonazoid microbubbles was tested for estimating right ventricular systolic and left ventricular diastolic pressures and compared with synchronously acquired catheter-based pressure measurements.

• ■ Mean errors in pressure measurements with the SHAPE technique were 1.6 mm Hg ± 1.5 [SD] for the right ventricular systolic pressure and were relatively higher for the left ventricular minimum and end-diastolic pressures (mean and median value of the errors ranging 5.3–8.4 mm Hg).

• ■ The results indicate that SHAPE with Sonazoid may have the potential to provide clinically relevant right ventricular systolic and left ventricular diastolic pressures noninvasively.

Introduction

Pressure measurements within the chambers of the heart provide essential information for the assessment and management of cardiac patients. The reference standard for obtaining intracardiac pressures is an invasive cardiac catheterization procedure, which precludes frequent pressure monitoring (1–3). Therefore, a noninvasive, accurate, and cost-effective approach for measuring intracardiac pressures will help patients in whom intracardiac pressure measurements are clinically indicated (1,2).

Subharmonic-aided pressure estimation (SHAPE) is a contrast-enhanced US technique that has shown promise for noninvasive pressure estimation in vitro and in vivo (4–10). The US contrast agents are gas-filled microbubbles (mean diameter, <8 μm) encapsulated by a lipid, protein, or polymer shell. With the insonation pressures greater than 100–150 kPa, these microbubbles act as nonlinear oscillators yielding energy components in the received echo signals at frequencies ranging from the subharmonic (half of transmit frequency) to higher harmonic and even ultraharmonic frequencies (11). The subharmonic signal amplitude decreases linearly with an increase in the ambient hydrostatic pressure when the insonation pressure range is 300–600 kPa (termed as the growth stage) (12). This inverse linear relationship between subharmonic signal amplitude in the growth stage and ambient pressure around the microbubbles exists for different US contrast agents and forms the basis of SHAPE (12–14). The SHAPE technique with Definity (Lantheus) microbubbles was reported as being encouragingly efficacious for obtaining intracardiac pressures noninvasively (7,10). However, of all the US contrast agents tested, the subharmonic amplitude of Sonazoid (GE HealthCare) microbubbles showed the highest sensitivity to detect ambient pressure changes (14 mm Hg per dB) (14). Sonazoid microbubbles consist of perfluorobutane gas encapsulated by a lipid membrane composed of hydrogenated egg phosphatidylserine sodium. Each vial consists of freeze-dried product and is reconstituted using 2 mL of sterile water. The reconstituted product contains 16 µL of perfluorobutane microbubbles (15).

The present study aimed to evaluate the use of Sonazoid microbubbles for intracardiac pressure estimation. Specifically, errors in right ventricular (RV) systolic and left ventricular (LV) diastolic pressures estimated using the SHAPE technique with Sonazoid microbubbles relative to standard measurements made with fluid-filled pressure catheters were computed. Additionally, solid-state high-fidelity pressure catheters limit the damping-associated artifacts in pressure measurements noticed with fluid-filled catheters and have been used in studies requiring high-quality pressure tracings (16–18). Therefore, in a separate group of participants, the LV pressure measurements were obtained using solid-state high-fidelity pressure catheters for comparison with the SHAPE technique.

Materials and Methods

Study Participants

Adults (>21 years of age) who were scheduled for a left or right heart catheterization procedure as part of their standard of care were prospectively enrolled in this institutional review board and U.S. Food and Drug Administration–approved study (Investigational New Drug [IND] no. 124465) from 2017 to 2020. The IND was obtained given that Sonazoid is not approved for use in the United States. The study was compliant with the Health Insurance Portability and Accountability Act. The inclusion and exclusion criteria are described in Table 1. Written informed consent was obtained from all participants. The study was registered with ClinicalTrials.gov (no. NCT03245255) prior to recruitment of the first participant. The total targeted enrollment for the study was 80 participants. GE HealthCare provided the Sonazoid microbubbles used in this study. All authors except one (K.D., who is an employee of Clarius Mobile Health) had control of inclusion of any data and information that might present a conflict of interest.

Table 1:

Participant Inclusion and Exclusion Criteria

SHAPE data acquisition.— Figure 1 depicts the data acquisition process in this study. All participants underwent a grayscale US examination prior to their catheterization procedure to determine the optimal acoustic window for visualizing the ventricles and aorta during the catheterization procedure. During the catheterization procedure, the participants were under conscious sedation (fentanyl and/or midazolam to effect). A 5F or 7F fluid-filled pressure catheter was advanced into the aorta and the LV or RV by the interventional cardiologist (M.P.S., D.F., N.R., P.W., A.V., with clinical experience ranging 5–48 years) to obtain systolic and diastolic pressures. Each Sonazoid vial was reconstituted per manufacturer recommendation using 2 mL of sterile water. Then, Sonazoid microbubbles from three vials (total volume, 6 mL) were withdrawn in a syringe and connected to a three-way stopcock for coinfusion with saline. Once the catheter was placed in the cardiac chamber, Sonazoid infusion (0.18 mL per hour per kg, along with saline at 120 mL per hour) was started using an existing intravenous line. A sonographer or physician verified the presence of Sonazoid microbubbles in the cardiac chambers using conventional grayscale imaging. A Doppler gate was placed in the LV or RV for SHAPE data acquisition. The Doppler gate enabled sampling from a specific area for pressure estimation; the real-time processing of the data acquired from the region of the Doppler gate was modified for SHAPE.

Figure 1:

Diagram shows the data acquisition process. LV = left ventricle, RV = right ventricle, SPU = short procedure unit.

The transmit parameters were 2.5-MHz transmit frequency and a square wave pulse shape, based on a previous study (19). To compensate for attenuation and phase aberration for each participant, an optimization algorithm was first initiated to determine the optimal incident acoustic output (IAO) for SHAPE data acquisition (19,20). In brief, this algorithm steps through and acquires subharmonic data at all IAO levels (coded as 0 to 15, 16 levels from 0.06 to 1.8 MPa peak to peak, 10 seconds per level) and determines the optimal IAO (12,14,19). After the optimization, data were acquired at the optimal IAO level (three times) synchronously with catheter-based pressure tracings for comparison. The time duration from initiating Sonazoid infusion to collecting SHAPE data from the LV or RV was approximately 10 minutes or less.

Data acquisition with high-fidelity catheter.— For a set of participants scheduled for left heart catheterization (n = 20), intracardiac pressures were also acquired using a solid-state high-fidelity Mikro-Cath (Millar) pressure catheter. In this subset, the SHAPE data were acquired synchronously with the Millar pressure catheter data for comparison, using the same protocol as described above. In this subset, the fluid-filled catheters were used for the catheterization procedure per the current standard of care. After the clinical procedure was completed, the Mikro-Cath pressure catheter was introduced for data acquisition.

Data acquisition for calculating the conversion factor.— Given that subharmonic data are in units of decibels (dB) and clinical pressure measurements are in units of millimeters of mercury (mm Hg), a conversion factor (in units of mm Hg per dB) is needed to transform the subharmonic data into pressure equivalence. Therefore, after the catheterization procedure was completed and while the patient was still on the procedure table, peripheral pressure measurements were acquired using a brachial cuff and SphygmoCor XCEL PWA (ATCOR) device. This device uses a transfer function with participant age and sex as input to derive the aortic pulse pressures from the brachial cuff–based pressure measurements (21). Using the aortic pulse pressures and subharmonic signal from the aorta, an individual conversion factor was calculated as the ratio of the aortic pulse pressure to the range of the subharmonic signals obtained from the aorta.

Data processing.— The pressure catheter recordings from the catheterization procedure were obtained from Mac-Lab (GE HealthCare). For the subharmonic data, the subharmonic signal amplitude range was first analyzed to see if signal variation was greater than 2.5 dB, which was considered the noise threshold (10). The subharmonic data at the optimal IAO level and the conversion factor were used to derive a SHAPE-based pressure waveform. For the resulting SHAPE-based LV pressure waveforms, the peak pressure was matched to the systolic LV pressure, and clinically relevant diastolic pressures (minimum and end diastolic) were noted. For the resultant SHAPE-based RV pressure waveforms, the diastolic pressure was matched to the diastolic RV pressure and systolic pressures were noted. Then, diastolic pressures from the LV and systolic pressures from the RV were compared with values obtained using the pressure catheter. The cardiologists (M.P.S., D.F., N.R., P.W., A.V.) acquiring the pressure catheter data were blinded to SHAPE data acquisition and measurements.

Statistical Analysis

The Pearson correlation coefficient was computed between the simultaneously acquired subharmonic signal and pressure catheter data for each participant. Bland-Altman analysis was used to compare pressure measurements between the SHAPE technique and the pressure catheter (22). Based on the distribution of the differences in cardiac pressure values obtained using the SHAPE technique and the clinical pressure catheter, either a two-tailed paired t test (normal distribution) or Wilcoxon signed rank test (nonnormal distribution) was used to compare the SHAPE technique to the reference standard. Bonferroni corrections were used for multiple comparisons. P < .05 was considered indicative of a statistically significant difference. All analyses were performed with Prism (version 9; GraphPad Software).

Results

Participant Characteristics

Participant enrollment for this study is shown in Figure 2. Table 2 presents the demographics of the enrolled participants. Of the 80 enrolled participants, research data were not acquired in 15 due to health concerns during the catheterization procedure, software and/or hardware issues related to the data acquisition process, patient scheduling and/or arrival of emergency cases, and withdrawn consent before the procedure. Participant health concerns included radial artery spasm, intravenous extravasation, and clinical decisions not to perform catheterization or suboptimal ulnar circulation impacting radial access. These health concerns may occur during the routine clinical cases or interventions and are not specifically related to the SHAPE research study. The hardware and/or software issues encountered were the US unit failing to power up and the SHAPE software failing to load. The most common reason for the cardiac catheterization procedure was an abnormal stress test (31 of 80 participants [39%]), followed by chest pain and/or shortness of breath (16 of 80 participants [20%]), angina and/or angina pectoris (six of 80 participants [8%]), heart failure and/or transplant (seven of 80 participants [9%]), follow-up for an LV assist device (six of 80 participants [8%]), and hypertension and other conditions. The participant weight–based dose of Sonazoid ranged from 7.2 to 27 mL per hour, and the total infusion time ranged between 5 and 10 minutes per participant. No adverse events were observed during Sonazoid infusion and SHAPE data acquisition. The optimal IAO level varied between participants, ranging from level 2 (0.3-MPa peak negative pressure or mechanical index [MI] of 0.19) to level 15 (0.6-MPa peak negative pressure or MI of 0.38), with a median value corresponding to level 10 (0.5-MPa peak negative pressure or MI of 0.31).

Figure 2:

Flowchart shows participant enrollment for this study. LV = left ventricle, RV right ventricle, SHAPE = subharmonic-aided pressure estimation.

Table 2:

Demographic Characteristics of Enrolled Participants

Correlation between Subharmonic Signal and Pressure Catheter Data

Figure 3 shows an example comparison of waveforms from the LV and RV obtained using the SHAPE technique and using the pressure catheter. The mean Pearson correlation coefficient values between the subharmonic waveforms and pressure catheter waveforms were −0.8 ± 0.1 [SD] for data obtained from the RV and LV. The Pearson correlation coefficient was used for comparing waveforms given the underlying data and pressure values used for comparisons followed a normal distribution. Note, the negative sign of the correlation coefficient implies the inverse linear relationship between the subharmonic amplitude from Sonazoid microbubbles and intracardiac pressures, as previously documented (14).

Figure 3:

Graphs show example (A) right and (B) left ventricular pressure waveforms at the optimal incident acoustic output with the subharmonic-aided pressure technique (dashed line) and with the fluid-filled pressure catheter (solid line) in two representative participants. The r² values were 0.67 and 0.71 for waveforms shown in (A) and (B), respectively. In (A), the arrow indicates the right ventricular systolic pressure on the fluid-filled pressure catheter waveform. In (B), the dashed and solid arrows indicate left ventricular end-diastolic and minimum diastolic pressures, respectively, on the fluid-filled pressure catheter waveform.

Comparison of SHAPE-based Pressures with Fluid-filled Pressure Catheter Measurements

RV data.— As shown in Figure 2, the SHAPE and pressure catheter data were obtained in 22 participants (of 30 enrolled participants for this group). From this set of 22 participants, the aortic data were not acquired in most, as the participants were scheduled for right heart catheterization only. Thus, we had insufficient data for calculating the conversion factor, which requires aortic data acquisition. Ideally, the SphygmoCor data would have provided the aortic pressures needed for conversion factor calculation. However, there have been documented differences between the aortic pressures reported by the SphygmoCor device and catheter-based aortic pressures, thus requiring an additional transfer function (21). Due to this, we could only process data from the participants with aortic pressure measurements available, which were only eight participants (eight of 22 [36.4%]) who were scheduled for left heart catheterization as well. Therefore, for these eight participants, errors in RV systolic pressures were computed (Table 3). We found no evidence of differences in the RV systolic pressures obtained with the SHAPE technique and those obtained with the fluid-filled pressure catheter (mean and median errors, 1.6 and 1.2 mm Hg, respectively; P = .85). A Bland-Altman analysis was performed, revealing an estimation bias of 0.15 mm Hg (Fig 4A).

Table 3:

Absolute Errors in Clinically Relevant Pressures Obtained Using the Fluid-filled Pressure Catheter during the Cardiac Catheterization Procedure and Estimated Using the Subharmonic-aided Pressure Estimation Technique

Figure 4:

(A–C) Bland-Altman plots show (A) right ventricular peak systolic pressures (n = 8) and left ventricular (B) minimum and (C) end-diastolic pressures (n = 17) obtained using the subharmonic-aided pressure estimation technique and the fluid-filled pressure catheter. Dotted lines represent the lower and upper limits of the 95% limits of agreement.

LV data.— To determine the errors in estimating LV diastolic pressures, data were acquired in 25 participants (of 30 enrolled participants for this group). In this group, error calculations were not performed in seven (of 25) participants because the subharmonic signal range was less than the noise threshold of 2.5 dB (10). Additionally for one participant, SphygmoCor data (required for conversion factor calculation) were not acquired per the protocol (ie, while the participant was still on the procedure table) due to an emergency cardiac case scheduled in the same laboratory immediately after the enrolled participant. For the remaining 17 of 25 participants (68%), we were able to compute the errors in LV diastolic pressures (Table 3). When compared with the catheter-based pressure measurements, pressures estimated using the SHAPE technique showed a statistically significant difference for the minimum diastolic pressures (mean and median errors, 8.4 and 6.5 mm Hg, respectively; P = .04), but not for the end-diastolic pressures (mean and median errors, 7.4 and 5.3 mm Hg, respectively; P = .09). Bland-Altman analysis showed a bias of 5.9 and 4.1 mm Hg for the minimum and end-diastolic errors, respectively (Fig 4B, 4C).

Comparison of SHAPE-based Pressures with High-Fidelity Solid-State Pressure Catheter Measurements

Figure 5 shows a comparison of waveforms from the LV for the fluid-filled and high-fidelity solid-state pressure catheters obtained in the same participant during the left heart catheterization procedure. Of 20 participants in this group, SHAPE and pressure catheter data (using the high-fidelity solid-state pressure catheter specifically for this study) were acquired in 18 participants (Fig 2). Data were not acquired in one participant due to the participant's condition during cardiac catheterization, and in another participant due to calibration issues with the high-fidelity solid-state pressure catheter system. Error calculations were performed in 10 of the 18 participants (Table 4) due to the subharmonic signal range being less than the noise threshold in the remaining participants (10). We found no evidence of differences for the minimum and end-diastolic errors obtained using the SHAPE technique compared with pressures obtained using the high-fidelity solid-state pressure catheter (median errors, 6.2 and 7.5 mm Hg for end-diastolic and minimum diastolic pressures; P ≥ .44). Bland-Altman analysis revealed a bias of 3.9 and 1.4 mm Hg for the minimum and end-diastolic pressures, respectively (Fig 6).

Figure 5:

Graph shows example left ventricular pressure waveforms from the fluid-filled (red) and high-fidelity solid-state (blue) pressure catheter obtained during cardiac catheterization in the same participant.

Table 4:

Errors in Clinically Relevant Pressures Obtained Using the High-Fidelity Solid-State Pressure Catheter during the Cardiac Catheterization Procedure and Estimated Using the Subharmonic-aided Pressure Estimation Technique

Figure 6:

Bland-Altman plots show left ventricular (A) minimum and (B) end-diastolic pressures (n = 10) obtained using the subharmonic-aided pressure estimation technique and the high-fidelity solid-state pressure catheter. Dotted lines represent the lower and upper limits of the 95% limits of agreement.

Discussion

The need to noninvasively obtain intracardiac pressures for monitoring cardiac patients has been echoed in literature for decades (1,2), and it has been documented that with the SHAPE technique, it may be possible to obtain in vivo clinical pressures noninvasively (4–7,9,10,12,14,19–21,23,24). Sonazoid was selected for this study based on a previous in vitro study, which concluded that Sonazoid was the US contrast agent most sensitive to changes in hydrostatic pressure (14). The present study evaluated the efficacy of the SHAPE technique by using a novel scanner interface that allowed for real-time optimization, which is known to be critical for accurate estimates (10,19). The mean and median errors between the SHAPE technique and the fluid-filled pressure catheter for the systolic RV pressures were 1.6 and 1.2 mm Hg, respectively (based on the absolute value of the differences), albeit based on a limited dataset. For the LV diastolic pressures, these errors were relatively high, with mean values of 8.4 and 7.4 mm Hg and median values of 6.5 and 5.3 mm Hg (based on the absolute value of the differences) for minimum and end-diastolic pressures, respectively. In the study group in which pressure measurements were obtained using the high-fidelity solid-state pressure catheter system, the mean and median errors were in the similar range (6.2–7.5 mm Hg).

There have been several approaches for pressure estimation using US contrast agents, but each of these had high error values (>15 mm Hg) or did not have a clinical component, making it impossible to determine safety and in vivo errors (12,25–29). A clinical pilot study investigating the efficacy of the SHAPE technique for measuring cardiac pressures showed that errors between SHAPE technique and the fluid-filled pressure catheter data can be as low as 2.6 mm Hg, with correlation coefficients as high as −0.9, but correlations as low as −0.3 were also noted (7). This wide range was due to the inability to determine the optimal IAO in real time. In the other clinical trial (using Definity microbubbles) that addressed the issues associated with the pilot study (7) by allowing determination of IAO in real time, relatively low errors were noted for RV systolic and LV diastolic pressures (mean errors ranging 1.7–2.9 mm Hg) (10). In the present study, the errors were higher than anticipated given that Sonazoid was most sensitive to changes in hydrostatic pressure when compared with other US contrast agents in an in vitro study (14). The higher-than-anticipated errors could be due to the variability between the optimized parameters used for in vitro (19) and clinical experiments. This suggests that different transmit pulse shaping options and transmit parameter settings (eg, transmit frequency, pulse shape) may need to be considered for intracardiac SHAPE with Sonazoid microbubbles based on preclinical or clinical data (8). Such transmit parameter optimization may result in increased subharmonic signal range, relative to the noise threshold of 2.5 dB, for pressure estimation. Another major source of this variability could be due to limitations imposed by the US scanning technique and the acoustic window available for scanning and data acquisition while participants are in the supine position during the cardiac catheterization procedure; this was not an issue in other documented preclinical studies with relatively lower reported errors (6,23,24). The lack of ability to move or turn participants to obtain the best acoustic windows for data acquisition (which is otherwise achievable in echocardiography) may have contributed to the differences in the magnitude of errors noted for RV and LV pressures. For LV diastolic pressures, errors less than 5 mm Hg are considered clinically acceptable (30). Finally, the errors obtained in this study with Sonazoid microbubbles were increased relative to errors noted in another study using Definity microbubbles (10), which suggests that the transmit parameters (transmit frequency, pulse shape) for cardiac SHAPE with Sonazoid microbubbles may need to be optimized further based on preclinical or clinical studies.

There were several limitations to this study. This includes the inability to move study participants during their clinical procedure to obtain a preferred acoustic window for contrast visualization and data acquisition for the SHAPE protocol. Also, the sample size was relatively limited. While we enrolled 80 participants for this study, the correlation between subharmonic signal and pressure catheter waveform data were computed in only 65 participants. Additionally, due to the differences noted between the aortic pressures acquired using the pressure catheter and those acquired using the SphygmoCor device (which provides central aortic pressures) (21), the conversion factor could not be calculated for most of the data collected from the RV. Only in cases where participants also underwent left heart catheterization (eight of 22) were the aortic pressures available to compute the conversion factor and estimate RV pressures using the SHAPE technique. While the SHAPE data and SphygmoCor data were acquired at different times and could be another source of error, a transfer function derived to convert SphygmoCor values to representative catheter pressure values may be applicable in future studies (21). Additionally, due to a low subharmonic signal range (<2.5 dB) in some participants, LV pressure estimates were obtained in 27 of 43 participants. In addition to the transmit parameter settings, the low subharmonic signal amplitude range for SHAPE could have been due to hardware limitations associated with the SonixTablet scanner; the available IAOs are limited to 16 discrete steps and, therefore, if the optimal IAO would be between these discrete steps, then the SHAPE errors may be relatively high, as shown previously (20). The lack of time available for SHAPE data acquisition during the standard-of-care cardiac catheterization was another limitation during data acquisition. Finally, given the clinical conditions of the enrolled participants, breath holding (as is required for echocardiographic examination) was difficult for many, which impacted the visualization of the ventricles. All of these limitations culminated in obtaining pressure measurements for comparisons in 35 of 65 participants where data were acquired. Therefore, additional studies that address issues related to the low subharmonic signal range, with relatively finer increments in the IAO, and incorporating this range check in the IAO optimization algorithm are needed.

In conclusion, RV and LV pressures estimated using the SHAPE technique with Sonazoid microbubbles were comparable with measurements obtained using the cardiac catheterization technique. The results indicate that the SHAPE technique with Sonazoid microbubbles may have the potential to noninvasively determine intracardiac pressures.