Assessment of preload responsiveness by variation in peak aortic flow velocity in suprasternal window: A novel approach

INTRODUCTION

Change in velocity time integral (VTI) measured with Doppler at the apical window in the left ventricular outflow tract (LVOT) has been the mainstay of assessing fluid responsiveness for more than a decade.[1] However, the inter-observer and intra-observer variabilities in VTI measurement have been found to be 14% and 11%, respectively.[2,3] These variabilities are similar to the fluid responsiveness thresholds of 10% and 15% used commonly, which may result in inadvertent fluid administration. A study, assessing aortic valve status in patients with aortic stenosis, found the aortic peak velocity to have less measurement error (3.2%) compared to LVOT-VTI (22%).[4] A few studies have demonstrated peak to peak respiratory variation in velocity (ΔVpeak) to more accurately predict fluid responsiveness than VTI at the descending aorta, carotid artery, and brachial artery.[5,6] The suprasternal window for measuring ΔVpeak in the descending aorta (∆VpeakdAo), with an added advantage of allowing intraoperative measurement without interfering with the surgical field, has also been suggested as a measure of fluid responsiveness.[5] However, this parameter has been utilised only in one paediatric study and there is no evidence available in mechanically ventilated adult patients. Therefore, we planned this study with an aim to determine the diagnostic performance of ∆ VpeakdAo at the suprasternal window and ΔVpeak in LVOT (ΔVpeakLVOT) at the apical window to assess fluid responsiveness using passive leg raise (PLR) manoeuvre.

METHODS

This observational study was performed from May 2024 to April 2025 after clearance from the ethics committee (vide approval number AIIMS/IEC/24/143, dated 15 March 2024) and prospective registration in the Clinical Trials Registry-India (vide registration number CTRI/2024/05/066669, dated 29/03/2024, accessible at http://ctri.nic.in). The principles of the Declaration of Helsinki (2013) and Good Clinical Practice Guidelines were followed during the conduct of the study. Out of 154 eligible patients, 140 adult patients between 18 and 65 years of age, posted for elective non-cardiac thoracic surgery cases consisting of lobectomy, wedge resection, pneumonectomy, decortication, and thymoma surgery, were included. The patients posted for emergency surgery, those with body mass index ≥30 kg/m², cardiovascular disease, poor echocardiographic window, history of myocardial infarction, patients on inotrope or vasodilator infusions, arrhythmias, negative modified Allen’s test, neurological condition with risk of an increase in intracranial pressure with PLR, abdominal compartment syndrome, and pregnant females were excluded. After attainment of a steady depth of anaesthesia, defined as 15 minutes after induction, baseline haemodynamic and echocardiographic measurements were obtained using a phased-array transducer (5S probe; Philips CX50).[2] All patients received set ventilation of 6 mL/kg with a peak end expiratory pressure (PEEP) of 5 cm H₂O. Parameters were recorded in two sequential steps. In the first step, echocardiographic measurements were obtained in supine position at apical and suprasternal windows. In the second step, stroke volume index (SVI) at the apical window and peak velocity at both sites were reassessed 1 minute after PLR to 45° to evaluate fluid responsiveness. At the apical window, LVOT-VTI and ΔVpeakLVOT were measured and averaged over three consecutive respiratory cycles. In the parasternal window, the diameter (D) of LVOT was measured. SVI and cardiac index were calculated from LVOT-VTI, LVOT diameter, and body surface area (BSA) using the formula: SVI = (LVOT-VTI × πD²/4)/BSA. Patients were classified as responders if SVI increased by more than 15% following PLR. From the suprasternal window, ΔVpeakdAo was measured over three respiratory cycles as ΔVpeakdAo = [(Vpeak_max − Vpeak_min)/{(Vpeak_max + Vpeak_min)/2}] × 100.

We hypothesised that ΔVpeakdAo measured at baseline would demonstrate good discriminative ability, with an area under the receiver operating characteristic (AUROC) of 0.70.[5] A total sample size of 140 patients was calculated with an anticipated 30% prevalence of fluid responsiveness to PLR and a confidence interval (CI) width of 0.20 for AUROC at 95% confidence level.[7] Baseline quantitative variables between responders and non-responders were compared using Student’s t-test or Mann–Whitney U test, wherever appropriate. Categorical variables were compared using Chi-square test/Fisher’s exact test. Receiver operating characteristic (ROC) curve analysis was performed to determine optimal threshold values and diagnostic performance of ΔVpeakdAo and ΔVpeakLVOT for predicting fluid responsiveness. A P value of < 0.05 was considered statistically significant.

RESULTS

Baseline parameters of the patients are summarised in Table 1. Among the study patients (n = 140), 24 (18%) were categorised as responders and 116 (82%) as non-responders. Volume expansion induced by PLR resulted in a significant increase in SVI in responders, from 34.03 [28.83–42.63] to 43.76 [38.1–55.44] mL/m² (P < 0.001), and a significant decrease in non-responders from 31.63 (27.34–34.23) to 29.17 (24.09–33.8) mL/m² (P = 0.002). ΔVpeakdAo was significantly higher in responders compared with non-responders (15.76% [12.95–18.51] vs. 13.02% [11.55–14.84], P = 0.019). However, there was no difference in ΔVpeakLVOT (18.92% [14.87–22.61] vs. 16.94% [14.1–20.32], P = 0.574 [Table 2].

Table 1.

Demographic data of the cohort, including responders and non-responders

Characteristics		Overall (n=140)		Responders (n=24)		Non-Responders (n=116)
Age (years), Mean±SD		41.71±16.38		41.25±11.57		41.811±7.25
Gender, n (%)
• Male		82 (58.6)		12 (50)		70(60.3)
Weight (kg), Mean±SD		55.63±10.66		54.75±7.05		55.81±11.28
BMI (kg/m2), Mean±SD		22.01±4.14		21.30±2.03		22.16±4.45
BSA (m2), Mean±SD		1.57±0.15		1.55±0.11		1.57±0.16
ASA grade, n (%)
• I • II • III		58 (41.4) 76 (54.3) 6 (4.3)		6 (25) 18 (75) 0 (0)		52 (44.8) 58 (50) 6 (5.2)
Comorbidities, n (%)
None		94 (67.1)		12 (50)		82 (70.7)
Smoker		24 (17.1)		12 (50)		12 (10.3)
Hypertension		0 (0.0)		0 (0.0)		0 (0.0)
Diabetes Mellitus		6 (4.3)		0 (0.0)		6 (5.2)
Pulmonary Tuberculosis		11 (7.9)		0 (0.0)		11 (9.5)
Epilepsy		5 (3.6)		0 (0.0)		5 (4.3)

ASA: American Society of Anesthesiologists; BMI: Body mass index; BSA: Body surface area; SD: Standard deviation

Table 2.

Haemodynamic and echocardiographic data recordings in responders and non-responders before and after PLR

Parameter		Overall patients		Responder (n=24)				Non-Responder (n=116)
				(before PLR)		After PLR (P)†		(before PLR) (P)^		After PLR (P)†
∆VpeakdAo (%)		13.2 [12.0-15.4]		15.76 [12.95-18.51]		16.38 [14.8-18.12] (0.040)		13.02 [11.55-14.84] (0.019)		13.29 [9.99-16.8] (0.732)
∆VpeakLVOT (%)		16.9 [14.1-21.7]		18.92 [14.87-22.61]		16.29 [14.88-18.04] (0.040)		16.94 [14.1-20.32] (0.574)		15.16 [13.51-16.88] (0.001)
LVOTVTI (cm)		15.3 [13.4-16.0]		12.55 [12.38-12.9]		16.45 [15.93-17.42] (0.001)		15.5 [14.3-16.18] (<0.001)		14.3 [13.1-15.62] (0.001)
HR (/minute)		101.0 [88.5-109.0]		106.5 [98.75-119.5]		94.5 [88-103.75] (0.001)		100.5 [87-108] (0.007)		82 [72-100] (0.001)
SBP (mmHg)		117.0 [110.0-137.0]		129.5 [106.75-159.75]		134 [123-142.75] (0.307)		117 [112-137] (0.466)		106 [102-131.75] (0.074)
DBP (mmHg)		74.5 [63.0-83.0]		74.5 [65-81.75]		77.5 [69.75-84] (1.000)		74.5 [63-83] (0.923)		68 [57-80] (0.790)
MAP (mmHg)		88.5 [76.0-95.0]		90.5 [79.5-104.25]		101 [87.75-110] (0.758)		88.5 [76-94] (0.446)		82.5 [71-101] (0.441)
dAoVTI (cm)		13.10 (11.00-16.50)		9.8 [8.9-11.93]		10.7 [9.6-12.5] (0.001)		13.7 [11.9-16.8] (0.001)		14.7 [13.3-16.2] (0.001)
Stroke Volume Index (mL/m2)		31.63 (27.34-35.81)		34.03 [28.83-42.63]		43.76 [38.1-55.44] (0.001)		31.63 [27.34-34.23] (0.059)		29.17 [24.09-33.8] (0.002)
CI (L/min/m2)		3.48 (2.65-3.71)		3.74 [3.49-4.39]		4.3 [3.95-5.19] (0.001)		3.31 [2.36-3.7] (<0.001)		2.54 [2.16-3.04] (0.001)

∆VpeakdAo: Vpeak respiratory variation in descending aorta; ∆VpeakLVOT: Vpeak respiratory variation in left ventricular outflow tract; CI: Cardiac index; dAoVTI: Velocity time integral of descending aorta; DBP: Diastolic blood pressure; HR: Heart rate; LVOTVTI: Velocity time integral of LVOT; MAP: Mean arterial pressure; SBP: Systolic blood pressure; (P)- P value; PLR: Passive leg raising. ^Wilcoxon–Mann–Whitney Test – Between-group comparisons at baseline; ^†Wilcoxon Signed-Rank Test – Within-group comparisons over time (Before PLR/After PLR); P<0.05 was considered significant

The AUROC for ΔVpeakdAo and ΔVpeakLVOT was 0.65 (95% CI: 0.55–0.76), and 0.54 (95% CI: 0.40–0.67), respectively. A baseline ΔVpeakdAo value >12.25% predicted responsiveness with a sensitivity of 100% (95% CI 86–100) and a specificity of 35.3% (95% CI 27–45). A baseline ΔVpeakLVOT value >21.59% predicted responsiveness with a sensitivity of 50% (95% CI: 29–71) and a specificity of 75% (95% CI: 66–83).

DISCUSSION

In this study, ΔVpeakdAo identified fluid responsiveness to PLR in mechanically ventilated adult patients with acceptable diagnostic power. A baseline ΔVpeakdAo value greater than 12.25% demonstrated potential utility as a cut-off for discriminating responders from non-responders to PLR. In contrast, ΔVpeakLVOT failed to predict changes in SVI, thereby limiting its utility for dynamic assessment of preload responsiveness.

Phased array probe in ultrasound is the most appropriate to assess Doppler flow measurements as they align more with actual flow. Only a few studies have assessed fluid responsiveness using a phased-array probe by measuring either ΔVpeak or ΔVTI in the descending aorta.[5,8,9] Another study evaluated fluid responsiveness using ΔVpeak in the carotid artery obtained in the supraclavicular view with a phased-array probe.[10] The cut-off value of 12.25% observed in our study agrees with the previously reported cut-off of 14%.[5] Doppler interrogation of ∆Vpeak at extracardiac sites such as the carotid artery was reported to have an acceptable (AUROC 0.65; CI: 0.55 to 0.76) diagnostic ability.[10] At the same time, some studies have failed to prove the diagnostic power of ∆Vpeak at other sites such as the carotid, descending aorta, and brachial artery.[6,11,12] Consequently, given the low positive predictive value, ΔVpeakdAo should be interpreted as an adjunctive parameter and integrated with overall clinical assessment rather than being used as a standalone diagnostic tool for fluid management decisions.

In the present study, ΔVpeakLVOT failed to predict response to PLR. This observation is in agreement with a prior research study in which ΔVpeakLVOT was not a reliable diagnostic marker of fluid responsiveness.[5] The limited performance of ΔVpeakLVOT may be attributed to variability in apical window positioning on the chest wall between measurements, displacement of the imaging plane during the respiratory cycle, and an increased likelihood of signal dropout. In contrast, the suprasternal window provides a more anatomically fixed and reproducible site of interrogation, with minimal influence from respiratory motion, resulting in greater measurement precision and reduced inter-observer variability compared with LVOT-based velocity measurements.[9]

We acknowledge a few limitations. Acquisition of the suprasternal view in adult patients was technically challenging and often required placement of a bolster beneath the shoulders to facilitate neck extension. Furthermore, peak velocity measurements obtained from the descending aorta exclude blood flow to the upper body, whereas LVOT-based measurements represent total cardiac output. The proportion of responders in our cohort was relatively low (18%) compared with the anticipated 30%, which may have affected the diagnostic accuracy estimates. Future studies with larger sample sizes, particularly in the intensive care unit or post-operative settings, are warranted to further validate these findings.

CONCLUSION

Assessment of ΔVpeakdAo from the suprasternal window provides an adjunctive means of predicting fluid responsiveness with acceptable accuracy.

Study data availability

De-identified data may be requested with reasonable justification from the authors (email to the corresponding author) and shall be shared after approval as per the authors’ institution’s policy.

Disclosure of use of artificial intelligence (AI)-assistive or generative tools

The AI tools or language models (LLMs) have not been utilised in the manuscript.

Author contributions

JK : Concepts, design, definition of intellectual content, literature search, conduct of cases, data acquisition, and manuscript preparation. AK: Concepts, design, definition of intellectual content, literature search, conduct of cases, data acquisition, data analysis, and manuscript preparation. MP, AD: Data analysis, manuscript editing, review, and approval.

Conflicts of interest

There are no conflicts of interest.

Funding Statement

Nil.