Effect of volume infusion on left atrial strain in acute circulatory failure

Background Left atrial strain (LAS) is a measure of atrial wall deformation during cardiac cycle and reflects atrial contribution to cardiovascular performance. Pathophysiological significance of LAS in critically ill patients with hemodynamic instability has never been explored. This study aimed at describing LAS and its variation during volume expansion and to assess the relationship between LAS components and fluid responsiveness. Methods This prospective observational study was performed in a French ICU and included patients with acute circulatory failure, for whom the treating physician decided to proceed to volume expansion (rapid infusion of 500 mL of crystalloid solution). Trans-thoracic echocardiography was performed before and after the fluid infusion. LAS analysis was performed offline. Fluid responsiveness was defined as an increase in velocity-time integral (VTI) of left ventricular outflow tract ≥ 10%. Results Thirty-eight patients were included in the final analysis. Seventeen (45%) patients were fluid responders. LAS analysis had a good feasibility and reproducibility. Overall, LAS was markedly reduced in all its components, with values of 19 [15 – 32], -9 [-19 – -7] and − 9 [-13 – -5] % for LAS reservoir (LASr), conduit (LAScd) and contraction (LASct), respectively. LASr, LAScd and LASct significantly increased during volume expansion in the entire population. Baseline value of LAS did not predict fluid responsiveness and the changes in LAS and VTI during volume expansion were not significantly correlated. Conclusions LAS is severely altered during acute circulatory failure. LAS components significantly increase during fluid administration, but cannot be used to predict or assess fluid responsiveness. Supplementary Information The online version contains supplementary material available at 10.1186/s13613-024-01274-6.


Background
Circulatory shock is present in up to one-third of patients admitted to the ICU and volume expansion represents the first-line therapy [1].Only 50% of critically ill patients increase their cardiac output after fluid infusion [2], a finding that prompts the accurate investigation of volume status in order to avoid useless fluid administration and deleterious consequences of fluid overload.Hemodynamic assessment of fluid loading should encompass the evaluation of left atrium (LA) performance and its contribution to cardiac function [3].
Left atrium is a complex and dynamic structure that significantly influences cardiovascular performance by actively coordinating with the left ventricle (LV) during the cardiac cycle [4].LA assumes different roles, functioning as a reservoir during ventricular systole, as a conduit during early ventricular diastole and as a pump that increases ventricular filling during late ventricular diastole [5].Structural and functional characteristics of LA reflect left ventricular diastolic function, as LA is exposed to LV filling pressure during diastole [6].
Gaps of knowledge in application of LAS in critically ill patients remain significant.The literature on the topic is scarce [19][20][21] and to date, the pathophysiological meaning of LAS in critically ill patients remains unknown.Few studies conducted in healthy subjects suggest that LAS is influenced by preload variations.LAS reservoir (LASr) is reduced in response to controlled reduction of cardiac loading; this includes tilting manoeuvre [22], Valsalva manoeuvre [23] and continuous positive airway pressure (CPAP) application [24].Conversely, load alteration during passive leg raising produces an increase in LASr [24].In patients with renal failure, LASr and LAS conduit (LAScd) were reduced after a preload reduction generated by the haemodialysis session, while LAS contraction (LASct) was unaffected [25,26].The effect of a fluid bolus on atrial strain in critically ill patients with acute circulatory failure has never been explored.The primary goal of this study was to describe the change in LAS during volume expansion in patients with acute circulatory failure.As secondary goal, we aimed to test if LAS or LAS variations during volume expansion accurately detect preload responsiveness.

Patients
This prospective observational study was conducted between June 2022 and September 2023 in the medical ICU of Henri-Mondor university hospital, Creteil, France.The study was approved by the ethics committee of the French Society of Intensive Care Medicine (Société de Réanimation de Langue Française, SRLF, 23-050).Because we routinely use echocardiography to assess the circulatory status of critically ill patients in our ICU, this technique was considered as a component of standard care and patient's consent was waived.Written and oral information about the study was given to the patients or families as per French law.
We included adult critically ill patients if the treating physician in charge decided to perform volume expansion because of the presence of acute circulatory failure.Acute circulatory failure was defined by at least one of the following signs: (a) hypotension (systolic arterial pressure ≤ 90 mmHg or decrease of more than 50 mmHg or mean arterial pressure ≤ 65 mmHg); (b) heart rate ≥ 100 bpm; (c) oliguria (diuresis ≤ 0.5 mL/kg/h) for more than 2 h; (d) lactate levels ≥ 2 mmol/L; (e) skin mottling.Exclusion criteria were supraventricular tachyarrhythmia or pacemaker rhythm at the time of inclusion, extracorporeal membrane oxygenation and poor image quality for LA strain analysis.Patients were included according to the availability of investigators and ultrasound systems.

Data collection
We prospectively collected demographic, clinical and biological information from electronic medical records.

Echocardiography
Transthoracic echocardiography (TTE) examination was performed by trained clinicians (competent in critical care echocardiography) in the supine position using the cardiac ultrasound probe (1.5-4.5 MHz, M5S-D) of a high-quality ultrasound system (GE Vivid S7 or E9 ultrasound system; GEMS, Buc, France).A standard echocardiographic protocol was used, performing all measurements according to current guidelines [27,28].For each patient, a single operator performed all TTE measurements.Echocardiographs were recorded and stored as DICOM files for offline analysis.All measurements were taken at end-expiration.A five-chamber apical view was used to record the velocity-time integral of the flow in the left ventricular outflow tract (LVOT-VTI), by averaging three measurements [29].Left atrial volumes and left atrial ejection fraction were determined by the application of biplane method.

Physiological significance of left atrial strain
Speckle-tracking echocardiography analysis of left atrial deformation allows measurement of LAS and provides information on all phases of atrial function: reservoir, conduit and contraction [28,30,31].Left atrial reservoir phase corresponds to pulmonary venous return during ventricular systole, when LA fills and stretches, generating a positive strain deflection.During the conduit phase, which starts with mitral valve opening and spans early ventricular diastole, LA passive emptying occurs, producing a decrease in LAS corresponding to the first negative deflection in LAS curve.Contraction phase occurs from the onset of atrial contraction and covers late ventricular diastole; it is characterised by atrial wall shortening, generating a second negative deflection on left atrial strain curve [32,33].

LAS analysis
LAS analysis was performed offline (EchoPAC, GE Healthcare) using an automated speckle tracking software with a LAS dedicated mode.LAS included three measurements: LASr, LAScd and LASct, reflecting atrial deformation during different phases of cardiac cycle (Fig. 1).LASr is a positive value while LAScd and LASct are negative values.LAS was measured using optimized apical-four-chamber (A4C) and apical-two-chamber (A2C) views, in order to visualize left atrial endocardium during the entire cardiac cycle.The region of interest (ROI) of the LA was defined by the endocardial border (inner contour of the LA wall) and the epicardial border (outer contour of the LA wall).The regions of interests (ROI) were generated automatically and manual adjustments were performed when necessary.Zero-baseline for obtaining LAS curves was set at end-ventricular diastole using R-R ECG gating.Atrial phases definition and LAS measurements were made in accordance with the European Association of CadioVascular Imaging / American Society of Echocardiography guidelines [30].We calculated mean LASr, LAScd and LASct by averaging the values recorded in four-and two-chamber views.
The software we used for the analysis of left atrial strain offers and automated measure of tracking quality.Moreover, we visually checked tracking quality by comparing the underlying atrial wall image loop with the superimposed tracking results.Lastly, we analysed the curves derived from that tracking.Only images where the LA wall could be properly traced were included.

Other measurements
We recorded the following standard hemodynamic data: heart rate, systolic arterial pressure, diastolic arterial pressure, mean arterial pressure, pulse pressure variation, and perfusion index.Respiratory variables, such as Fig. 1 Left atrial strain measurement *A4C indicates apical-four-chamber view; A2C, apical-two-chamber view; LASr, left atrial strain reservoir; LAScd, left atrial strain conduit; LASct left atrial strain contraction, pre-A, pre-atrial contraction respiratory rate, tidal volume and positive end expiratory pressure were also collected.

Study protocol
At baseline, a first set of hemodynamic and TTE data was collected.Then, a fluid bolus of 500 mL of a crystalloid solution was performed in less than 15 min [34].A second set of measurements was obtained immediately after volume expansion.Throughout the study period, the ventilator settings and the dose of sedatives and vasoactive drugs were left unchanged.Fluid responsiveness was defined as an increase of 10% or more in LVOT-VTI after volume expansion [35,36].

Reproducibility
We assessed intra-observer and inter-observer variability for LAS analysis in a sample of 10 patients that were randomly selected.Inter-observer variability was evaluated by asking two operators (M.C. and F.B.) to perform LAS analysis of the same exams blinded to each other results.Intra-observer variability was evaluated by asking one investigator (M.C.) to repeat LAS analysis with at least 2-week interval, blinded to the results of the first analysis.The repeated analyses were performed on the same pre-selected loops and cardiac cycles.Inter-and intraobserver variability were assessed by intraclass correlation coefficients.

Statistical analysis
Sample size calculation was based on LASr value.We calculated that a sample size of at least 29 patients would have a 90% power to detect a 5% improvement in LASr after fluid bolus, considering a baseline LASr of 20% with a standard deviation of 9%, based on previous studies evaluating LAS in critically ill patients [19,21].Considering the difficulty in obtaining good-quality images in critically ill patients, we planned to enrol at least 40 patients.Data are expressed as mean ± standard deviation or median [interquartile range], as appropriate.Normality of continuous variables was assessed with the Shapiro-Wilk test.Comparisons between before and after fluid administration were assessed through a paired Student's t test or a Wilcoxon test, as appropriate.Comparisons between fluid responders and fluid non-responders were assessed through a two-sample Student's t test or a Mann-Whitney U test, as appropriate.Because left atrial strain has been shown to reflect left ventricular filling pressure in patients with heart failure [8,12], we tested its association with traditional Doppler indices of abnormal diastolic function [37] by using bivariate correlation analysis and summarizing the results in a correlation matrix (corrplot package within the R environment).Correlations were tested using the Spearman method with Benjamini-Hochberg correction to control the false discovery rate at the 0.05 level.Receiver-operating characteristic curves (with 95% confidence interval) were built for the prediction of fluid responsiveness using the baseline value of LAS.Correlations were quantified by the Spearman coefficient.A p-value of less than 0.05 was considered statistically significant.

Patient characteristics
Fifty-three patients were assessed.Ten patients were excluded (including five patients with non-sinus rhythm at the time of inclusion, five patients for whom data were lost for when transferring echocardiographs from ultrasound machine to hard disk), and five patients had inadequate atrial wall visualisation for LAS assessment, leaving 38 patients for analysis (feasibility of 88% [38/43], Fig. 2).All patients were included once.

Hemodynamic variables
Overall, fluid loading induced an increase in arterial pressure, cardiac output and pulse oximetry, and a decrease in pulse pressure variation (Table 2).Seventeen (45%) patients were fluid responders.There was no significant difference in hemodynamic characteristics at baseline between fluid responders and non-responders (Table 2).

Echocardiographic variables
Table 2 displays the echocardiographic parameters at baseline.Mean LAS was feasible in 36 patients.The intra-and inter-observer variability was good for all components of LAS, with intraclass correlation coefficients all above 0.8 (Table S1).All LAS components were markedly reduced (twice lower) at baseline as compared to reported reference values.All LAS components significantly increased during volume expansion (Table 2).Within groups, LASr increased both in fluid responders and non-responders, while LAScd increased only in non-responders and LASct increased only in responders (as also did left ventricle global longitudinal strain) (Table 3, Figure S1-S3).None of the components of LAS at baseline predicted fluid responsiveness (Table 2, Figure S4), and the change in LAS was not associated with the change in LVOT-VTI during fluid loading (Figure S5).The correlations between LAS variation al LV global longitudinal strain variation in responders and nonresponders are shown in Figure S6.There was no difference between fluid responders and non-responders concerning parameters of left ventricular and left atrial structure, except for lower values of left ventricle volumes in the former group (Table 2).Indices of diastolic function were similar between groups, apart from higher values of E-wave deceleration time in responders.In the correlation matrix, most indices of increased left ventricular filling pressure were not associated with LAS, nor in the entire cohort nor in non-responder patients.(Figure S7).

Discussion
We herein report the first study on LAS during fluid loading in the critically-ill with the following main findings: (i) LAS was feasible with good reproducibility in critically ill patients with circulatory failure; (ii) all three components LAS were severely altered at baseline and markedly increased with fluid administration overall; (iii) the baseline values of LAS did not predict fluid responsiveness, and the change in LAS was not correlated to the change in cardiac ejection.

LAS in critically ill patients
Several studies have reported normal reference values of LAS derived from large populations of healthy adult subjects [38,39]

LAS and volume expansion
Our study examined the effect of fluid-induced changes in cardiac preload on LAS.Some evidence exists that LAS may be preload dependent.Previous small studies [20,[22][23][24] enrolling mainly healthy subjects, have shown an   association between preload alteration and LASr.LAS variation during an increase of cardiac preload has only been evaluated by Gottfridsson et al. [24], who documented an increase in LASr during a passive leg raising manoeuvre in healthy young individuals.Our study confirms these findings, as we described a major increase in LASr during rapid fluid administration.The conduit and contraction phases of left atrial function have been poorly studied during controlled load alterations.
There is some evidence [23][24][25] indicating that LASct is relatively unaffected by preload variations.However, the Valsalva manoeuvre and continuous positive airway pressure application used in studies by Gottfridsson et al. [23,24] shall not produce significant changes in preload in all subjects.In our study, we found a statistically significant increase in LASr and LASct with fluid loading in responders, along with an increase in LV systolic strain.These finding are consistent with cardiovascular physiology.Mechanistically, LASr is coupled to LV longitudinal shortening since the two chambers are anatomically connected, and in systole, the LV exerts a direct stretching effect on the atrium [5].In fluid responsive patients, ventricular contractile function improves and this could be an important contributing mechanism to the rise in LASr in this setting.The absence of significant correlation between LAS variation and LV longitudinal strain seems to contradict these assumptions and needs further explorations.Similar to LV, LA pump function is theoretically determined by preload (Frank-Starling mechanism), a fact that may explain our finding of increased LASct with fluid loading in responders.Supporting this hypothesis, we found a tendency for an increase in left atrial ejection fraction during volume expansion in fluid responders, although not statistically significant.Ünlü et al. [25] found that in patients with end-stage renal disease, LASr and LAScd declined during the haemodialysis session.In our study, both LASr and LAScd increased in non-responders during fluid loading.These observations may suggest the role of LA pressure, but warrant further research, inasmuch as the absolute changes in LAScd values were small in our cohort.

Fluid intolerance
Left atrial strain has been used in cardiology patients for identifying elevated left ventricular filling pressure [8] and for providing a more accurate categorization of diastolic dysfunction than do conventional echocardiographic variables [40].Similarly, LA strain measure could be used to detect fluid intolerance in critically ill patients receiving a fluid bolus, outperforming traditional echocardiographic indices of diastolic dysfunction.
The majority of studied patients did not show echocardiographic criteria of diastolic dysfunction according to EACVI criteria [37], neither before nor after fluid administration.More specifically, we found that only one patient could be classified as having elevated left ventricular filling pressure and three patients indetermined LV filling pressure at baseline.The absence of significant correlation between LA strain and conventional indices of diastolic dysfunction is consistent with the characteristics of included patients and timing of echocardiographic examination: we enrolled patients early after the onset of acute circulatory failure and before substantial fluid resuscitation.
In our study we didn't find any difference in the three components of LAS (LASr, LAScd and LASct) between fluid responders and non-responders patients, nor in conventional echocardiographic criteria of diastolic dysfunction (except for deceleration time of E-wave).Moreover, LASr increased during volume expansion by the same rate in fluid responders and non-responders.
In conclusion, according to our results, LAS doesn't appear to add clues to the identification of patients that would not benefit from fluid administration, and further studies are warranted to assess its usefulness to detect fluid intolerance.

Fluid responsiveness
Over the last 20 years, traditional static markers of preload responsiveness, notably central venous pressure [41], have been shown to be unreliable and several dynamic tests have been developed (e.g., pulse pressure  [8].In our study, we showed that a static value of LAS could not detect fluid responsiveness.In this perspective, our finding that LAS components cannot predict fluid responsiveness is consistent with previous literature on static indicators of cardiac preload [36].

Limitations
Our study has several limitations.First, this is a singlecentre study with a cohort of small size; in addition, inclusion rate was slow.Therefore, the generalizability of our results is questionable.Second, we included a general population of critically ill patients needing volume expansion as the clinician acumen and we might have missed the most severe patients; nevertheless, we wanted our study to be pragmatic and thus to describe left atrial strain change during volume expansion in patients that actually receive intravenous resuscitation fluid in clinical practice [42][43][44].Unfortunately, we didn't record the specific trigger for fluid administration in our cohort of patients and we cannot exclude that LAS physiology could be different in specific subgroups.Third, the measurement of LAS requires an experienced operator and the recording of adequate apical views.However, in our study we found a good feasibility (only 5/43 patients were excluded because of inadequate ultrasound imaging) and reproducibility of LAS parameters.Nevertheless, TTE image quality in critically ill patients is often suboptimal and we did not investigate to which extent echogenicity could impact the precision of LAS analysis.Finally, we used change in VTI as a surrogate of increase in stroke volume to define fluid responsiveness.However, echocardiography is widely used nowadays in ICU to track changes in cardiac output and the relative changes in VTI have been validated in this setting [29].

Conclusion
All components of LAS were severely altered during acute circulatory failure and significantly increased during fluid administration.However, neither their baseline value nor their variation can be used as markers of fluid responsiveness.

Table 1
Patients characteristics at baseline

Table 3
[35]dynamic and echocardiographic variables during fluid expansion in responders and non-responders , stroke volume variation, passive leg raising, end-expiratory occlusion test)[35]. .LAS is assumed to reflect at least in part left ventricular end-diastolic pressure, with lower values of LAS being markers of elevated left ventricular filling pressure * Data are expressed as number (percentage) or median (interquartile range).).HR indicates Heart Rate; bpm, beats per minute; SAP, Systolic Arterial Pressure; DAP, Diastolic Arterial Pressure; MAP, Mean Arterial Pressure; LVEF, Left ventricular ejection fraction; LVEDV, left ventricular end diastolic volume; LVESV, left ventricular end systolic volume; LVOT-VTI, Left ventricle outflow tract -velocity-time integral; CO, Cardiac Output; GLS-LV, global longitudinal strain -left ventricle; E-wave early diastole; A, A-wave, atrial contraction; DTE, deceleration time E-wave; E' lat, lateral early diastolic annulus relaxation velocity; LA, left atrium; LAVI, left atrial volume index; LAEF, left atrial ejection fraction; LAS, left atrial strain; LASr, LAS reservoir; LAScd, LAS conduit; LASct, LAS contraction variation