Association between Preoperative Dynamic Measures of Vascular Load and Postoperative Hemodynamics in Patients with Chronic Thromboembolic Pulmonary Hypertension after Pulmonary Thromboendarterectomy

To the Editor:

Chronic thromboembolic pulmonary hypertension (CTEPH) is caused by a heterogeneous distribution of pulmonary arterial thromboemboli (1–3) that undergo fibrotic transformation and result in increased pulmonary artery pressure (PAP) and pulmonary vascular resistance (PVR). The increased right ventricular (RV) afterload leads to RV hypertrophy, dilation, and eventual dysfunction/failure. However, among types of pulmonary hypertension, CTEPH is unique, as it is potentially curable via pulmonary thromboendarterectomy (PTE) (4, 5). Previous work has demonstrated that patients with residual pulmonary hypertension after PTE (PVR > 500 dyn · s/cm⁵) are at 10 times higher risk of death (10.3% vs. 0.9%) than patients with lower PVR (<500 dyn · s/cm⁵) (6). However, current clinical markers are not able to robustly identify these patients preoperatively.

Ventricular contractility, quantified as end-systolic elastance (Ees), and vascular load (pulmonary elastance [Epa] and pulmonary vascular compliance [Cpa]) are gold-standard measures of cardiovascular function (7–9). However, estimates of these parameters have typically required invasive and complex research procedures. Recent efforts have led to the estimation of these parameters from conventional clinical measures, and ventricular–vascular coupling has been shown to predict freedom from transplant in patients with pulmonary hypertension (10). Given the CTEPH-associated changes in both the pulmonary vasculature and RV performance, we hypothesized that clinically available measures of ventricular contractility and vascular load would improve the prediction of poor postoperative outcomes in patients undergoing PTE, and we evaluated these parameters in a pilot study.

Methods

Patient population

Thirty-seven consecutive patients with CTEPH (age 52.3 ± 15.3 yr; 62.2% female) undergoing PTE with preoperative right heart catheterization and echocardiographic data as well as postoperative hemodynamics and clinical history were retrospectively reviewed. The study was approved by our Institutional Review Board for Human Experimentation.

Hemodynamic and echocardiographic measures

Preoperatively, hemodynamic variables were measured during a pulmonary artery catheterization and included measurement of pulmonary artery systolic pressure (PAPs), diastolic pressure (PAPd), mean pressure (PAPm), and pulse pressure (PAPp) as well as cardiac output via thermodilution (COtd), stroke volume (SV; defined as COtd/heart rate), pulmonary capillary wedge pressure (PCWP), and pulmonary vascular resistance (PVR; defined as [PAPm − PCWP]/COtd). Catheterization was performed 9.3 ± 15.3 days before PTE surgery (range 2–84 d; median 6 d).

Postoperatively, pulmonary artery catheter hemodynamic measurements were obtained at the time of discharge from the intensive care unit (ICU) (2.5 ± 2.0 d; range 1–9 d; median 2 d). PCWP was not acquired in the ICU, and central venous pressure was used to estimate PVR. Poor hemodynamic outcome was defined on the basis of postoperative PAPm > 30 mm Hg or PVR > 300 dyn · s/cm⁵ (11). These outcomes were analyzed separately and as a composite endpoint. In addition, elevated total pulmonary resistance (TPR) (>800 dyn · s/cm⁵) was also analyzed as an individual outcome as well as part of a composite endpoint (with PAPm > 30 mm Hg).

Estimation of dynamic measures

Cpa was calculated from pulmonary artery catheterization parameters as the SV/PAPp. The single-beat method was used to estimate RV Ees, Epa, and RV–PA coupling (Ees/Epa). To do so, first, an average representative RV pressure waveform was created from digitized raw pressure waveforms from cardiac catheterization (12), as shown in Figure 1A. As shown in Figure 1B, Ees was calculated as Pmax − RVPes/SV, in which Pmax is the theoretical maximal RV pressure (calculated from nonlinear modeling of the early systolic and diastolic portions of the RV pressure curve) and RVPes is the end-systolic RV pressure estimated as PAPm (10, 13–15). Epa was calculated as PAPm/SV (10, 13). This leads to an estimate of Epa that is not the inverse of Cpa because of the use of PAPm and PAPp, respectively. RV–PA coupling was defined as Ees/Epa, which was estimated as Pmax/PAPm – 1 (10).

Figure 1.

(A) Right ventricular pressure waveforms were digitized, and maximum ventricular pressure was estimated on the basis of nonlinear fitting of the early systolic and diastolic portions of the curve. (B) Maximum ventricular pressure allowed for estimation of Ees. RV–pulmonary artery coupling was derived using estimation of arterial elastance, as shown. Cpa = pulmonary compliance; ED = end-diastolic point; EDV = end-diastolic volume; Ees = end-systolic ventricular elastance; Epa = pulmonary elastance; ES = end-systolic point; ESV = end-systolic volume; PAPm = mean pulmonary artery pressure; PAPp = pulmonary artery pulse pressure; Pmax = maximum ventricular pressure; RV = right ventricle; RVPed = end-diastolic right ventricular pressure; RVPes = end-systolic right ventricular pressure; SV = stroke volume.

Statistical analysis

Statistical analyses were performed in MATLAB (The MathWorks). To identify variables associated with poor postoperative outcomes, L1-penalized maximum likelihood (LASSO) multivariate regression, a method for building parsimonious regression models, was used with cross-fold validation (folds = 5). Preoperative conventional hemodynamic (PAPs, PAPd, PAPm, PAPp, PVR, and COtd) as well as dynamic measures (Cpa, Epa, Ees, and Ees/Epa) were eligible for inclusion in the logistic models for high postoperative PAPm (>30 mm Hg), PVR (>300 dyn · s/cm⁵), the PAPm + PVR composite endpoint (either PAPm > 30 mm Hg or PVR > 300 dyn · s/cm⁵), TPR (>800 dyn · s/cm⁵), and the PAPm + TPR composite endpoint.

Results

Change in hemodynamics after PTE surgery

Patient characteristics and surgical findings are presented in Table 1. None in the cohort died or had prolonged post-PTE hemodynamic instability. Consistent with prior findings of PTE, we observed improvements in hemodynamics (6).

Table 1.

Patient demographics and PTE surgical parameters

	Results
Patient demographics
Sex, % female	62.2
Age, yr	52.3 ± 15.3
BMI	29.7 ± 8.4
Surgical parameters
UCSD surgical classification of disease level
Left lung	2.5 ± 0.8
Right lung	2.2 ± 0.7
Cardiopulmonary bypass time, min	248.9 ± 31.0
Total circulatory arrest time, min	47.9 ± 21.7
Cooling time, min	90.1 ± 12.4
Rewarming time, min	122.6 ± 15.9
Days until discharge	12.1 ± 8.8
Need for inotropic support	6/37 (16.2%)
NYHA class	2.8 ± 0.6
Class I	2/37 (5.4%)
Class II	5/37 (13.4%)
Class III	28/37 (73.0%)
Class IV	3/37 (8.1%)

Definition of abbreviations: BMI = body mass index; NYHA = New York Heart Association; PTE = pulmonary thromboendarterectomy; UCSD = University of California, San Diego.

Table 2.

Changes in hemodynamics and echocardiographic measures after PTE

	Before PTE	After PTE	P Value
PAPs, mm Hg	77.2 ± 22.2	40.8 ± 15	<0.001
PAPd, mm Hg	26.9 ± 9.5	15.5 ± 7	<0.001
PAPp, mm Hg	50.3 ± 15.5	24.6 ± 10.4	<0.001
PAPm, (mm Hg)	45.6 ± 13.6	25 ± 8.9	<0.001
PVR, dyn · s/cm5	713.1 ± 445.4	250.2 ± 140.5	<0.001
COtd, L/min	4.4 ± 1.1	5.3 ± 1.2	<0.001

Definition of abbreviations: COtd = cardiac output via thermodilution; PAPd = pulmonary artery diastolic pressure; PAPm = mean pulmonary artery pressure; PAPp = pulmonary artery pulse pressure; PAPs = pulmonary artery systolic pressure; PTE = pulmonary thromboendarterectomy; PVR = pulmonary vascular resistance.

Clinical measures of patient outcomes

Poor clinical outcomes were defined based on hemodynamic variables. After PTE, 4/37 (11%) patients had PAPm > 30 mm Hg (mean, 40.7 ± 11.7 mm Hg), whereas 33/37 patients had PAPm ≤ 30 mm Hg (mean, 22.0 ± 10.3 mm Hg). PVR > 300 dyn · s/cm⁵ occurred in 9/37 (24%) patients (mean, 434.2 ± 166.7 dyn · s/cm⁵), whereas 28/37 had PVR ≤ 300 dyn · s/cm⁵ (mean, 191.1 ± 94.6 dyn · s/cm⁵). Eleven of 37 patients had the composite endpoint of PAPm > 30 mm Hg or PVR > 300 dyn · s/cm⁵, whereas only 2/37 patients had both PAPm > 30 mm Hg and PVR > 300.

TPR > 800 dyn · s/cm⁵ occurred in 12/37 (32%) patients, and 13/37 (35%) patients had the composite endpoint of PAPm > 30 mm Hg or TPR > 800 dyn · s/cm⁵.

Prediction of post-PTE PAPm and PVR

LASSO regression of post-PTE PAPm > 30 mm Hg resulted in a model with three parameters (PAPd, Epa, and Ees) with 90.6% accuracy and 96.6% specificity. LASSO regression of post-PTE PVR > 300 dyn · s/cm⁵ resulted in a model with two parameters (PVR and Cpa) with 72% model accuracy and 88% specificity. Predicting the PAPm + PVR composite endpoint resulted in a model with four parameters (PAPd, PAPm, PVR, and Cpa) with 91% accuracy and 93% specificity.

Predicting post-PTE TPR > 800 dyn · s/cm⁵ resulted in a model with two parameters (PVR and Epa) with 71.9% accuracy and 81.0% specificity. Predicting the PAPm + TPR composite endpoint resulted in a model with the same two parameters (PVR and Epa) but lower performance (62.5% accuracy and 71.4% specificity).

Discussion

In this retrospective pilot study of 37 consecutive patients with CTEPH, we found that conventional (PAPd, PAPm, and PVR) and dynamic (Epa, Ees, and Cpa) pre-PTE measures were predictive of post-PTE hemodynamics (PAPm, PVR, and TPR). The removal of endovascular obstructions in the pulmonary vascular bed via PTE is expected to correlate with static measures such as PAPm and PVR (which are clinically used to assess disease burden) as well as dynamic measures. Specifically, the obstructions can lead to changes in pulmonary artery elastance/compliance, and the right ventricle undergoes remodeling over time. Our study finds that measures of both pulmonary artery dynamics parameters and RV functional measures were also important in predicting postoperative changes.

Separate models for postoperative PAPm and PVR led to different predictive variables, likely because of the LASSO approach and differences in the strength of correlations between measures. However, the composite endpoint approach indicated that the parameters can be combined to achieve high diagnostic performance.

PTE provides patients with CTEPH a potentially curative treatment that, when performed at experienced centers, has low operative mortality. However, the increasing diagnosis and treatment of more distal disease subtypes, the advent of minimally invasive balloon pulmonary angioplasty, and the approval of riociguat for nonoperative or postoperative patients have all led to renewed interested in preoperative parameters, which can predict treatment outcomes.

Clinical outcomes, such as ICU duration, need for inotropes, and length of stay, were not included in this study. A larger study that can control for other factors, such as operative complications and comorbidities, is merited to further evaluate the use of these parameters. Furthermore, enrollment of sequential patients in a retrospective pilot study did not control for the percentage of patients with poor outcomes. Future work will need to focus on a larger cohort or potentially use a case-control design to help increase the outcomes of interest. Lastly, long-term follow-up information was not available as part of the patient record. As a result, measures of hemodynamic change were limited to time to discharge.

Conclusions

Dynamic measures of ventricular function and vascular loading (estimated via Ees, Epa, and Cpa derived from clinical measures) were correlated with post-PTE PAPm, PVR, and TPR.