A novel speckle-tracking echocardiography parameter assessing left ventricular afterload

Abstract

Background

Left ventricular stroke work index (LVSWI) and afterload-related cardiac performance (ACP) consider left ventricular (LV) afterload and could be better prognosticators in septic cardiomyopathy. However, their invasive nature prevents their routine clinical applications. This study aimed to investigate (1) whether a proposed speckle-tracking echocardiography parameter, Pressure-Strain Product (PSP), can non-invasively predict catheter-based LVSWI, ACP and serum lactate in an ovine model of septic cardiomyopathy; and (2) whether PSP can distinguish the sub-phenotypes of acute respiratory distress syndrome (ARDS) with or without sepsis-like conditions.

Methods

Sixteen sheep with ARDS were randomly assigned to either (1) sepsis-like (n = 8) or (2) non-sepsis-like (n = 8) group. Each ARDS and sepsis-like condition was induced by intravenous infusion of oleic acid and lipopolysaccharide, respectively. Pulmonary artery catheter-based LVSWI (the product of stroke work index, mean arterial pressure and.0136), ACP (the percentage of cardiac output measured to cardiac output predicted as normal) and serum lactate were measured simultaneously with transthoracic echocardiography. Two PSP indices were calculated by multiplying the mean arterial blood pressure and either global circumferential strain (PSPcirc) or radial strain (PSPrad).

Results

PSPcirc showed a significant correlation with LVSWI (r² =.66, p <.001) and ACP (r² =.82, p <.001) in the sepsis-like group. Although PSP could not distinguish subphenotypes, PSPcirc predicted LVSWI (AUC.86) and ACP (AUC.88), and PSPrad predicted serum lactate (AUC.75) better than LV ejection fraction, global circumferential and radial strain.

Conclusions

A novel PSP has the potential to non-invasively predict catheter-based LVSWI and ACP, and was associated with serum lactate in septic cardiomyopathy.

1. Introduction

Septic patients can develop cardiac dysfunction called septic cardiomyopathy (SCM). SCM does not have standardized diagnostic criteria, but it is often diagnosed as an acute cardiac disorder caused by sepsis, which is reversible and can be restored within 7−10 days.^1–3 The prevalence of SCM in septic patients ranges from 10% to 70%² and is associated with two to three times higher mortality rates, up to 70%−90%, when compared with the mortality in septic shock without SCM.^3,4 Hence, the accurate diagnosis of SCM is paramount for the timely management of the accompanying cardiac dysfunction.

Currently used conventional cardiac parameters such as cardiac output (CO) and echocardiographic ejection fraction (EF) often fail to detect cardiac dysfunction in the hyperdynamic stage of SCM due to the decreased left ventricular (LV) afterload and following hyperdynamic cardiac contraction, where CO and EF remain in normal or even high range.^2,3,5,6 Speckle-tracking echocardiography (STE) strain, especially global longitudinal strain (GLS), can assess cardiac function more sensitively than conventional EF and has been reported to demonstrate a higher association with mortality in patients with severe sepsis or septic shock.^4,7–9 Nevertheless, this isolated strain parameter can still be affected by LV afterload^10–12 and thus it remains controversial whether GLS is an ideal prognostic predictor in sepsis.⁴

Left ventricular stroke work index (LVSWI) and afterload-related cardiac performance (ACP) consider LV afterload and could be better prognosticators in SCM.^5,13 However, their invasive nature using catheter prevents their routine clinical applications.

In this study, we propose a novel STE strain parameter, Pressure-Strain Product (PSP), a product of mean arterial pressure (MAP) and end-systolic strain based on GCS (PSPcirc) or GRS (PSPrad). Since PSP incorporates both LV afterload (i.e. MAP) and LV contractility similar to LVSWI, PSP has the potential to correlate with LVSWI. We also hypothesise that PSP could be associated with ACP as a load-considering parameter, and serum lactate for predicting prognosis in sepsis.^14–16 The aims of this study are to investigate whether (1) PSP can non-invasively predict catheter-based LVSWI and ACP, as well as serum lactate in sepsis-like condition; and (2) whether PSP can distinguish the disease severity of acute respiratory distress syndrome (ARDS) with or without sepsis-like condition.

2. Methods

Reporting of the study conforms to broad EQUATOR guidelines.¹⁷

2.1. Study design

This study was conducted as a secondary analysis of an exploration of ovine ARDS phenotypes in a blinded, randomized, controlled preclinical trial as previously described.¹⁸ The animal experiments were conducted at the Queensland University of Technology (QUT) Medical Engineering Facility (MERF) in Brisbane, Australia. Animal ethics was approved by QUT Office of Research Ethics and Integrity (No. 18-606), in accordance with the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes and the Animal care and Protection Act 2001, QLD.

The study included 16 female, non-pregnant Merino−Dorset crossbreed ewes, aged 1−3 years, with a mean weight of 50 ± 3 kg. The sheep were randomly assigned to either the (1) Septic group (S group): lipopolysaccharide (LPS)-induced sepsis-like condition (n = 8) following oleic acid (OA)-induced ARDS, or the (2) non-septic group (NS group): OA-induced ARDS (n = 8) without LPS. Study outcomes were catheter-based cardiac parameters (i.e. LVSWI and ACP), proposed novel STE parameters (i.e. PSPcirc and PSPrad), conventional cardiac parameters (i.e. LV EF, GCS and GRS), and serum lactate.

2.2. Animal model

The animal model was previously described by Millar et al.,¹⁸ and the detailed methods are provided in the supplemental document. Animals were sedated with midazolam, fentanyl and ketamine through a four-lumen central venous line (CVL, Arrow Int.) inserted into the jugular vein, and were mechanically ventilated (Galileo 5, Hamilton Medical) via a cuffed 8.5 mm endotracheal tube (Mallinckrodt). Thereafter, tracheostomy using a size 8–9 mm tracheostomy tube (Portex, Smiths Medical) was performed. The animals were placed in a prone position and ventilated using a lung-protective ventilation protocol as per the EXPRESS trial.¹⁹ Haemodynamic parameters were monitored by electrocardiogram (ECG), femoral artery PiCCO catheter (PULSION Medical Systems, Getinge) and by a pulmonary artery catheter (Swan−Ganz catheter, Edwards Lifesciences LLC) inserted through an 8.5 Fr right jugular venous sheath (Edwards Lifesciences). After instrumentation procedures were completed and 1-h resting period, ARDS as per the official American Thoracic Society workshop report criteria²⁰ was induced as follows¹⁸: (a) NS group: sequential administration of oleic acid (O1008; Sigma-Aldrich) in subsequent 0.03 mL/kg doses intravenously (IV) through the distal end of the CVL until a PaO₂/FiO₂ (PF) ratio of <150 mmHg was reached; and (b) S group: aforementioned IV oleic acid until a PF ratio <150, followed by 0.5 μg/kg of LPS (E. coli O55:B5, Sigma-Aldrich) dissolved in 50 mL of normal saline, and infused over 1 h. The animals were then observed for 48 h. A schematic of the experimental timeline is provided in Figure S1.

2.3. Haemodynamic monitoring

ECG waveform, heart rate (HR) and invasive arterial blood pressure were continuously monitored throughout the experiments. Pulmonary artery pressure, central venous pressure, stroke volume (SV), cardiac index (CI) and systemic vascular resistance index (SVRI) were continuously monitored via a pulmonary artery catheter.

LVSWI is generally calculated by the formula [1]²¹:

$$\text{SVI}\times \text{(MAP-PCWP)}\times 0.0136,$$ (1)

where SVI is stroke volume index and PCWP is pulmonary capillary wedge pressure (Figure 1A). We deemed the value of PCWP as zero since PCWP could not necessarily be obtained in all animals due to anatomical peculiarities of sheep pulmonary arterial vasculature.

Figure 1.

Schematic images of LVSWI (A) and PSP (B). Y axis shows LV pressure, and X axis shows LV volume (A) and strain (B). The area surrounded by blue and green diagonal shows estimated LV stroke work and PSP, respectively. GCS (absolute value) and GRS can be used as a peak systolic strain for the calculation of PSPcirc and PSPrad, respectively. EDP, end-diastolic pressure; GCS, global circumferential strain; GRS, global radial strain; LV, left ventricular; LVSWI, left ventricular stroke work index; MAP, mean arterial pressure; PSP, Pressure-Strain Product; SVI, stroke volume index.

ACP is calculated using the formula [2]:

$$\text{ACP}=\frac{\text{CO}\text{measured}}{\text{CO}\text{predicted}}\times 100(\%),$$ (2)

whereas CO_predicted is calculated using the formula [3]:

$${\text{CO}}_{\text{predicted=}}560.68\times {\left\{\frac{80\times \text{(MAP-CVP)}}{\text{CO}}\right\}}^{-0.645},$$ (3)

where MAP is mean arterial pressure and CVP is central venous pressure.

2.4. Echocardiography

Transthoracic echocardiography was performed through a Vivid-i ultrasound machine (GE Vingmed Ultrasound) with a 3S-RS transducer or an IE-33 (Philips) with an X5-1 transducer from the left side of the chest whilst the animal was in a prone position at baseline and at 2, 5, 24 and 48 h following the model creation. Standard echocar-diographic parameters including LV wall thickness, LV diameters, LVEF (Teichholz) and LV fractional area change were measured based on the parasternal short axis (PSAX) view at the mid-papillary muscle level. All images were transferred to a separate workstation as DICOM data and analysed offline by an experienced cardiologist using the vendor-independent STE strain analysis platform (TomTec imaging Systems GMBH). STE parameters, global circumferential strain (GCS), global radial strain (GRS) and early diastolic strain rate (eDSR) of the endocardial layer were assessed based on three different levels of PSAX view (i.e. the levels of mitral valve, mid-pupillary muscle and apex), and each value was calculated as a mean. A rotation was automatically evaluated based on the two different levels (i.e. the levels of mitral valve and apex). Semi-automated measurement was applied to appropriate echo-loops. Speckle tracking was visually assessed for tracking accuracy and the end-diastolic and end-systolic timing markers were manually adjusted if required. The frame rate of 55/s for IE33 and 72/s for Vivid-i was applied to the strain assessment. The concept of PSP is illustrated in Figure 1B and its value was calculated by using the formula; MAP × GCS (absolute) for PSPcirc or MAP × GRS for PSPrad (Figure 2). Since the difference of clinical implication or utility between GCS and GRS is not yet fully elucidated,^22–24 both PSPcirc and PSPrad were investigated in this study.

Figure 2. An example of image for the measurement of PSPcirc and PSPrad.

(A) LV short axis view at mid-papillary level showing the myocardial layer (dark blue). (B) Two strain curves: endo-myocardial GCS (pink) and GRS (light blue). Each value at the end-systolic phase (eS, highlighted in yellow) was used for the calculation of PSPcirc and PSPrad. (C) Echocardiographic parameters including myocardial and endocardial GCS, EDA, ESA, FAC, GRS, and delta-rotation as the average value from the basal, mid-papillary, and apical levels. (D) Formula for calculating PSPcirc and PSPrad using endocardial GCS and GRS, respectively. EDA, end-diastolic area; ESA, end-systolic area; FAC, fractional area change; GCS, global circumferential strain; GRS, global radial strain; MAP, mean arterial pressure; PSPcirc, pressure strain product based on GCS; PSPrad, pressure strain product based on GRS. ROT, rotation; SD-CS, standard deviation of circumferential strain; SD-RS, standard deviation of radial strain.

2.5. Blood samples, cardiac biomarkers, inflammatory cytokines and histological assessment

The detailed methodology is described in the supplemental method.

2.6. Statistical analysis

Our primary analysis estimated correlations between proposed novel STE parameters (i.e. PSPcirc and PSPrad) and the catheter-based parameters (i.e. LVSWI and ACP), as well as serum lactate. Correlations were inferred from linear regression analysis performed on S and NS groups independently. In addition, the predictability of cardiac parameters for previously known prognosticators (LVSWI, ACP and serum lactate) was analysed by logistic regression. Dependent variables were LVSWI, ACP and serum lactate. Since the diagnostic criteria of these values for sepsis or SCM do not exist, they were divided into two groups by mean or median to represent lower versus higher levels. Independent variables include PSP-circ, PSPrad, LV EF, GCS and GRS. Analysis outcomes were reported as receiver operating characteristics (ROC), alongside estimates for the area under the curve (AUC), sensitivity and specificity.

Our secondary analysis investigated whether PSP could distinguish the sub phenotypes of ARDS with or without sepsis-like conditions. For this purpose, logistic regression analysis was conducted to assess the performance for prediction of NS group. The dependent variable was membership to the NS group. Independent variables include PSPcirc, PSPrad, LV EF, GCS, GRS, LVSWI and ACP. Results were reported as odds, with corresponding 95% confidence intervals and p-values.

Two-way analysis of variance (ANOVA) was conducted to compare cardiac parameters (e.g. blood pressure, biomarkers and echocardiographic parameters) between S and NS group over time. The data until 24 h rather than 48 h were analysed due to the low number of survivals at 48 h.

Inter-observer variability for the echocardiographic parameters was analysed by two experienced cardiologists using random cases. Intra-observer variability for the same parameters using the same animals mentioned above was also analysed 2 years after the initial analysis. All hypothesis testing was two-tailed and p <.05 was considered statistically significant. All statistical analysis was performed with SPSS for Mac 29.0 (SPSS Inc.).

3. Results

3.1. Study population

A total of 16 female sheep (8 for S group and 8 for NS group) were included and analysed. Baseline characteristics and survival rate in both groups are described in Table S2. No significant differences were observed between groups at baseline.

3.2. Linear regression analysis

1. LVSWI and other cardiac parameters

In a linear regression analysis, moderate but statistically significant associations were observed between LVSWI and PSP in both S (PSPcirc: r² =.663, p <.001, PSPrad: r² =.589, p <.001, Figure 3) and NS (PSPcirc: r² =.543, p <.001, PSPrad: r² =.663, p <.001, Figure S2) groups.

Figure 3. Linear regression analysis between LVSWI and other cardiac parameters in S group.

Scatter plot graphs between LVSWI and (A) CI, (B) LVEF, (C) GCS, (D) GRS, (E) PSPcirc and (F) PSPrad were described. Each assessment included the valid cases from the timepoint of baseline to 48 h following model creation. CI, cardiac index; GCS, global circumferential strain; GRS, global radial strain; LVEF, left ventricular ejection fraction; LVSWI, left ventricular stroke work index; PSPcirc, pressure strain product based on GCS; PSPrad, pressure strain product based on GRS.

2. ACP and other cardiac parameters

In a linear regression analysis, moderate but statistically significant associations were observed between ACP and PSP in both S (PSPcirc: r² =.817, p <.001, PSPrad: r² =.620, p <.001, Figure 4) and NS (PSPcirc: r² =.558, p <.001, PSPrad: r² =.470, p <.001, Figure S3) groups.

Figure 4. Linear regression analysis between ACP and other cardiac parameters in S group.

Scatter plot graphs between ACP and (A) CI, (B) LVEF, (C) GCS, (D) GRS, (E) PSPcirc and (F) PSPrad were described. Each assessment included the valid cases from the timepoint of baseline to 48 h following model creation. ACP, afterload-related cardiac performance; CI, cardiac index; GCS, global circumferential strain; GRS, global radial strain; LVEF, left ventricular ejection fraction; LVSWI, left ventricular stroke work index; PSPcirc, pressure strain product based on GCS; PSPrad, pressure strain product based on GRS.

3. Serum lactate level and cardiac parameters

In a linear regression analysis, weak but statistically significant associations were observed between serum lactate and PSP in both S (PSPcirc: r² =.156, p =.023, PSPrad: r² =.166, p =.021, Figure 5) and NS (PSPcirc: r² =.204, p =.008, PSPrad: r² =.185, p =.013, Figure S4) groups.

Figure 5. Linear regression analysis between serum lactate and cardiac parameters in S group.

Scatter plot graphs between ACP and (A) CI, (B) LVEF, (C) GCS, (D) GRS, (E) PSPcirc and (F) PSPrad are described. Each assessment included the valid cases from the timepoint of baseline to 48 h following model creation. CI, cardiac index; GCS, global circumferential strain; GRS, global radial strain; LVEF, left ventricular ejection fraction; LVSWI, left ventricular stroke work index; PSPcirc, pressure strain product based on GCS; PSPrad, pressure strain product based on GRS.

3.3. ROC curve analysis

ROC curve analysis was conducted to investigate whether echocardiographic parameters can predict previously known prognosticators in sepsis or SCM (i.e. LVSWI,¹³ ACP⁵ and serum lactate^15,16). These three values in all animals (n = 16) including all timepoints (from baseline to T48) were divided into binary group (i.e. higher or lower group) based on median (34 g-min/m² for LVSWI and 77% for ACP) or mean (2.0 mmol/L for serum lactate) as per distribution. Performance for prediction of these three prognosticators was greater for PSP (circ or rad) than other echocardiographic parameters, LVEF, GCS and GRS (Figure 6).

Figure 6.

ROC curve analysis to investigate the performance of echocardiographic parameters for prediction of previously known prognostic values. Prognostic values include LVSWI, ACP and lactate. Echocardiographic parameters were compared among LVEF, GCS, GRS, PSPcirc and PSPrad. Each prognostic value (including all animals and all timepoints) was divided into either higher or lower value based on mean or median as per distribution. ACP, afterload-related cardiac performance; CI, cardiac index; GCS, global circumferential strain; GRS, global radial strain; LVEF, left ventricular ejection fraction; LVSWI, left ventricular stroke work index; PSPcirc, pressure strain product based on GCS; PSPrad, pressure strain product based on GRS.

3.4. Binary (S vs. NS) logistic regression analysis in cardiac parameters

In the logistic regression analysis for prediction of NS group using cardiac parameters, none of the invasive and non-invasive measurements including PSP could show significant odds for a 1SD increase in each parameter (Table S3).

3.5. Haemodynamics

All the haemodynamic parameters significantly changed over time (p <.001, Figure S5). Between-group differences in haemodynamic parameters were not statistically significant except for MAP (p =.027, partial η² =.208) (Figure S5).

All eight sheep in the S group met criteria of septic shock as defined in The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3) (requiring vasopressors to maintain the MAP over 65 mmHg and having a serum lactate level of 2.0 mmol/L or more)²⁵ during observation.

3.6. Biomarkers

A significant increase in high-sensitivity troponin I (p =.007) and serum lactate (p =.002) over time was observed, while no significant difference was observed between the groups (Figure S6). The serum interleukin 6 (IL-6), which has been reported to be a myocardial depressant in SCM,²⁶ significantly increased over time (p <.001). S group demonstrated significantly higher IL-6 when compared to the NS group (Figure S6).

3.7. Echocardiographic parameters

Representative echocardiographic images are shown in Figure S7. Conventional and STE strain parameters during observation are described in Figures S8 and S9, respectively. Overall, LV EF (p =.001, Figure S8), GCS (p =.004, Figure S9), and GRS (p <.001, Figure S9) significantly changed over time, while no significant differences in conventional and STE echocardiographic parameters were observed between the S and NS group (Figures S8 and S9).

3.8. Histological assessment

Contraction band necrosis and myocytolysis were more severe in S group compared to NS group, although non statistically significant (Table S4). Representative histological findings are described in Figure S10.

3.9. LVSWI, ACP and PSP

LVSWI, ACP and PSP transiently dropped at T2, followed by a gradual recovery but not to baseline levels. No significant differences were observed in these parameters between S and NS group (Figure S11)

3.10. Inter- and intra-observer correlation analysis in echocardiographic parameters

Seven cases were randomly chosen for the inter- and intra-observer correlation analysis. Analysis was done by two experienced cardiologists. There was a good to excellent inter- and intra-observer correlation among echocardio-graphic parameters including PSPcirc (ICC.941, p =.002, and ICC.978, p <.001, respectively), and PSPrad (ICC.876, p =.002, and ICC.863, p =.018, respectively) (Table S5).

4. Discussion

In the present study, we found that non-invasively measured PSP had a stronger linear association with pulmonary artery catheter-derived LVSWI and ACP than conventional parameters in an ovine model of SCM-like heart. PSP was also associated with serum lactate. The novelty of PSP compared to traditional strain parameters could be its nature to consider LV afterload and the potential to non-invasively predict catheter-based prognosticators (i.e. LVSWI and ACP). PSP has the potential to reflect accurate cardiac pump function adjusted by blood pressure and can be associated with the well-known prognostic predictor, serum lactate level.

4.1. Appropriateness of the ovine model for the SCM-like LV dysfunction under ARDS

All 16 sheep met the criteria for severe ARDS (PaO₂/FiO₂ ratio <150 mmHg) in large animal models as well as criteria in histological evidence of lung tissue injury.²⁰ S group met the criteria of septic shock as mentioned in ‘Haemo-dynamics’ in Section 3.

In terms of cardiac dysfunction, CI significantly dropped from baseline at T2, but gradually recovered to the baseline levels thereafter. This reversible cardiac dysfunction does not contradict the feature of SCM (Figure S5). Although the standardized criteria for the histological features of SCM are yet to be established,^27–29 some authors have reported that SCM mainly consists of myocardial oedema,^27,28 and a discrepancy exists at this point in our study. This could be because the cytotoxic pattern of lesions may be more predominant than oedematous changes in our study, or our result might be affected by the timing of tissue sampling. In any case, the appropriateness of the model for the SCM can be improved in future studies.

4.2. Correlation between PSP and LV load-considering parameters (LVSWI and ACP)

PSP significantly correlated with pulmonary artery catheter-derived LVSWI and ACP as an LV load-considering parameter. Other echocardiographic parameters (LV EF, GCS and GRS) were less correlated with LVSWI and ACP compared to PSP. This can be because PSP, LVSWI and ACP rely mainly on MAP. Linear regression analysis also showed a significant association between LVSWI and ACP in the S group (r² =.828, p <.001) and in the NS group (r² =.721, p <.001) (Figure S13). This indicates that there is a significant association among these three (PSP, LVSWI and ACP) load-considering parameters by incorporating the value of MAP in their formulae.

4.3. PSP and disease severity

PSP could not discriminate two ARDS sub-phenotypes with or without sepsis-like conditions. This could be because not only the S group but also the NS group had cardiac injuries and subsequent haemodynamic deterioration by a single OA injection without LPS. However, only PSP (both PSPcirc and PSPrad) among other echocardiographic parameters showed a weak but statistically significant association with the serum lactate level (Figures 5 and 6), which could reflect disease severity and inadequate CO in sepsis.^15,16 PSP also showed an improved performance in predicting the value of LVSWI and ACP (Figure 6), which are also deemed as better prognostic predictors in sepsis¹³ and SCM.⁵ This indicates that PSP can be a better prognosticator than other cardiac parameters such as LV EF and conventional STE parameters (GCS and GRS). Whether PSPcirc or PSPrad is more useful in clinical sepsis cohort needs to be further examined.

4.4. Clinical implication of PSP

PSP is expected to reflect LV stroke work as an LV pump function by quantifying the ability to generate blood pressure as well as LV contractility. A higher value of PSP means good pump function and vice versa. Given that blood pressure is an important component as a target of treatment in septic shock,³⁰ it is reasonable to incorporate the value of blood pressure in cardiac assessment for the prediction of prognosis in sepsis. As observed for LVSWI in a previous study,¹³ lower PSP could potentially predict higher mortality in a sepsis cohort. This hypothesis needs to be further investigated in future clinical studies.

4.5. Potential advantages of PSP as a cardiac marker in sepsis

As suggested in a previous study,¹³ LVSWI rather than common echocardiographic parameters such as LV EF could better predict the prognosis for cardiac intensive care patients including sepsis. This could be because LVSWI incorporates measures of both LV preload and afterload, and could potentially overcome the limitation of load dependency in LV EF. In this sense, PSP, as a potential surrogate of LVSWI, could be useful in assessing cardiac function to detect SCM.

One of the potential advantages of PSP could be its broad applicability for any severity of sepsis. In the most severe case of sepsis (i.e. septic shock), conventional cardiac parameters can stay in the normal range due to the decreased LV afterload (i.e. pseudo-normalisation), but PSP will decrease as the MAP will drop drastically. This corrected cardiac performance for LV afterload can be sensitive to detect even hyperdynamic SCM. In contrast, in the early stages of sepsis before shock, PSP would also decrease as PSP accounts for myocardial strain which can detect subtle cardiac injury that cannot be caught by conventional LV EF or stroke volume.^4,7–9 Hence, PSP might be useful regardless of the stage of sepsis.

Another potential advantage of PSP compared to invasive LVSWI/ACP or echocardiography-based LVSWI/ACP could be the simplicity of the formula as it consists of just two non-invasive parameters: MAP and end-systolic strain. Recent newer ultrasound machines embed STE analysis software,⁷ and thus PSP could be obtained bedside in a non-invasive manner. In addition, PSP can be calculated in a vendor-independent fashion, which may overcome the limitation of the similar strain-based and vendor-dependent parameter, myocardial work.³¹ Although multivendor variability in STE strain remains a limitation, this can be overcome by using vendor-independent software.³²

4.6. Limitations

There are several limitations to this study. First, the relatively small sample size reduces the certainty of our conclusions. Second, PCWP was deemed as zero in our formula when calculating catheter-based LVSWI because of the complexity of obtaining PCWP in our ovine model. Therefore, in the case of SCM with elevated LV end-diastolic pressure, our result could have less validity. Applying MAP subtracted by the estimated LV filling pressure (e.g. trans mitral E/e′ velocity ratio¹³) instead of MAP alone may yield better agreement with LVSWI. Lastly, we applied GCS and GRS rather than GLS for the calculation of PSP due to the anatomical constraints in our animal model. GLS-based PSP could be preferable considering that GLS is more commonly used in clinical settings with higher sensitivity and reproducibility compared to GCS and GRS.³³ However, the optional applicability of GCS and GRS in the prediction of LVSWI could be an advantage in the case that adequate apical views are not obtainable in intensive care equipped with ventilator or other mechanical support. In addition, given that LV circumferential fibre is more related to right ventricular (RV) pressure than longitudinal fibre,³³ PSP based on GCS may be a better cardiac parameter than GLS by reflecting both LV and partial RV function. This could be advantageous especially in ARDS where pulmonary hypertension and following RV dysfunction can be observed.³⁴

5. Conclusion

A novel STE parameter, PSP based on GCS and GRS, has the potential to non-invasively predict catheter-based LVSWI, ACP and serum lactate in SCM-like conditions. The utility of PSP as a prognostic predictor in clinical sepsis needs to be further investigated.