Comparison of subclavian vein to inferior vena cava collapsibility by ultrasound in acute heart failure: A pilot study

Abstract Background Management of acute decompensated heart failure (ADHF) requires accurate assessment of relative intravascular volume, which may be technically challenging. Inferior vena cava (IVC) collapsibility with respiration reflects intravascular volume and right atrial pressure (RAP). Subclavian vein (SCV) collapsibility may provide an alternative. Hypothesis The purpose of this study was to examine the relationship between SCV collapsibility index (CI) and IVC CI in ADHF. Methods This was a prospective study of non‐ventilated patients with ADHF who had paired IVC and SCV ultrasound assessments. As SCV CI is highly position‐dependent, measurements were performed supine at 30–45°. Results Thirty‐three patients were included with 36 encounters. The sample size was adequately powered for receiver‐operator characteristic (ROC) analysis. SCV CI correlated with IVC CI during relaxed breathing (R = .65, n = 36, p < .001) and forced inhalation (R = .47, n = 36, p = .0036). SCV CI < 22% and >33% corresponded to IVC CI < 20% and >50% suggesting hypervolemia (sensitivity/specificity: 72%) and hypovolemia (sensitivity/specificity: 78%), respectively. Moderate to severe tricuspid regurgitation (TR) compared to less than moderate TR was associated with lower SCV CI (medians: 12.4% vs. 25.3%, p = .022) and IVC CI (medians: 9.6% vs. 35.6%, p = .0012). SCV CI and IVC CI were not significantly different among chronic kidney disease stages. Conclusion In non‐ventilated ADHF, SCV CI at 30–45° correlates with paired IVC CI, and may provide an alternative to IVC CI for assessment of relative intravascular volume, which may facilitate clinical management. Moderate to severe TR decreases SCV CI and IVC CI and may result in overestimation of relative intravascular volume.


| INTRODUCTION
Early, accurate, and ongoing intravascular volume assessment in patients admitted with heart failure is required to provide effective goal-directed therapy, minimize adverse outcomes of inappropriate, ineffective, or excessive diuresis, and improve survival. Although 65% of nonedematous patients with chronic heart failure are intravascularly hypervolemic, 30% are euvolemic, and 5% are hypovolemic. 1 Assessment of relative intravascular volume by physical examination may be inaccurate, 1,2 and N-terminal prohormone of brain natriuretic peptide (NT-proBNP) is nonspecific for volume overload. 1 The American Society of Echocardiography supports the use of ultrasound assessment of IVC maximum diameter (IVC max ) and IVC collapsibility index (CI) with respiration to estimate right atrial pressure (RAP), 3,4 with recent validation. 5 IVC CI measurements have reasonable sensitivity (84%-96%) and specificity (37%-93%) for diagnosis of congestive heart failure, 6 and may help guide diuresis and heart failure therapy to improve patient outcomes. 7 The goal of volume management is to improve cardiac output by optimizing relative intravascular volume. The gold standard for assessing volume responsiveness has been defined as an increase in cardiac output of ≥10% in response to volume administration. 2,8,9 Dynamic parameters that take into account respiratory variation, such as IVC CI 10,11 and subclavian vein (SCV) CI, 12 have the best sensitivity and specificity for estimating relative intravascular volume and predicting response to volume administration. These dynamic parameters are objective measures of the jugular venous waveform, with increased respiratory variation suggesting intravascular hypovolemia or volume responsiveness.
Ultrasound measurements of IVC CI have been used to assess relative intravascular volume overload in patients with congestive heart failure 6,7,[13][14][15] and/or renal failure 16,17 and to predict which patients may benefit from volume removal. An IVC CI of <20% was related to the ability to remove volume by ultrafiltration 16 and/or increase cardiac output after volume removal 17 in renal failure. In patients with shortness of breath due to various etiologies, IVC CI was greater in nonheart failure (46%) than in heart failure (9.6%) patients (p < .0001). 13 At an IVC CI cutoff of 15%, sensitivity and specificity for the diagnosis of congestive heart failure were 92% and 84%, respectively. 13 IVC diameter and variation with respiration may be altered by changes in relative intravascular volume as well as by moderate to severe tricuspid regurgitation (TR). 15,[18][19][20] SCV CI has been studied as an alternative assessment of relative intravascular volume, as IVC imaging can be technically challenging. 21 SCV CI measured with the patient reclining at a 30°angle predicted volume responsiveness in mechanically ventilated critically ill patients. 12 SCV CI measured in patients reclining at 30-45°correlated well with subcostal IVC CI in patients undergoing echocardiography, 22 in patients treated in the surgical intensive care unit 21 and in those hospitalized with acute and/or chronic renal failure. 23 SCV CI has not previously been compared to IVC CI in patients hospitalized with acute congestive heart failure.
The purpose of this study is to compare point-of-care ultrasound of SCV to IVC as an assessment of relative intravascular volume in patients hospitalized with acute exacerbation of heart failure.   Figure 1). The probe was placed inferior to the lateral border of the right clavicle, aligned in the deltopectoral groove along the axis of the right arm to obtain a transverse view of the SCV at the junction of the proximal axillary vein ( Figure 1A). 21,23,24 The IVC was imaged via the subcostal window in a longitudinal plane using the phased array transducer ( Figure 1A). For both IVC and SCV, maximum (D max ) and minimum (D min ) diameters were measured with both relaxed breathing and forced inspiration or "sniff." A frame-by-frame analysis of greyscale (B-Mode) images was performed to identify D max and D min , which were then measured using the digital calipers (GE Healthcare) ( Figures 1B,C). 25 Encounters were excluded if either the SCV or IVC were not adequately visualized.

| METHODS
After the initial assessment, each patient's chart was reviewed by the first author for further data collection (Table 1).
Data were presented as median ± interquartile ranges, as data were not normally distributed. Pearson's correlation coefficient was calculated to assess the relationship between SCV CI and IVC CI, with both relaxed breathing and forced inhalation.  14,16,[28][29][30] In these publications, an RAP of ≥20 mmHg was associated with IVC CI < 20% in 88% of encounters, and RAP of <5 mmHg was associated with IVC CI > 50% in 80% of encounters. 16 The sample size was based on the time available to recruit patients for this project, which is a pilot study. Subsequently, power analysis for the data obtained was used to determine whether the study was sufficiently powered. 31,32

| RESULTS
As shown in Figure 2, 33 patients were included, with 36 unique patient encounters. Three patients were examined and scanned twice during the same admission.
Patient characteristics are shown in Table 1.
The paired SCV and IVC collapsibility indices were significantly correlated during relaxed breathing (R = .65, n = 36, p < .001, Figure 3A) and during forced inhalation (R = .47, n = 36, p = .0036, Figure 3B). Bland-Altman analysis indicated that there was no systematic bias between SCV and IVC collapsibility.  Due to the inferior correlation with forced inhalation, only relaxed breathing was considered for subsequent analyses.
Based on prior studies, IVC CI > 50% and <20% have been shown to correspond to RAP by right heart catheterization of <5 mmHg and >20 mmHg, respectively. 16,23 Sensitivity and specificity of SCV CI cutoffs predicting these specific IVC CI cutoffs are shown in Figure 4. At all possible discrimination thresholds for SCV CI, sensitivities and specificities for whether SCV CI below the cutoff predicts IVC CI < 20% are shown in Figure 4A, AND whether SCV CI greater than the cutoff predicts IVC CI > 50% are shown in Figure 4B. SCV CI of <22% best predicted IVC CI < 20% (unlikely hypovolemia), with sensitivity/specificity of 72%. SCV CI > 33% best predicted IVC CI > 50% (unlikely hypervolemia), with sensitivity/specificity of 78%. As shown in Figure 5A, for the first encounter used for each patient, moderate to severe TR was associated with significantly lower median SCV CI (12.4%) compared to mild or no TR (25.3%, p = .022). Likewise, median IVC CI ( Figure 5B) was significantly lower with moderate to severe TR (9.6%) compared to mild or no TR (35.6%, p = .0012).
As shown in Figure 5C, median SCV max was significantly larger with moderate to severe TR (0.89 cm) compared to less than moderate TR (0.58 cm, p = .00097). Median IVC max with moderate to severe TR was also significantly larger (2.4 cm) than with less than moderate TR (2.1 cm, p = .026) ( Figure 5D).
The presence or absence of right-sided intracardiac devices was not associated with differences in SCV CI (p = .67 Mann-Whitney) or IVC CI (p = .54 Mann-Whitney). F I G U R E 4 Sensitivity and specificity for SCV CI cutoffs, as predictors for IVC CI < 20% or >50%, with relaxed breathing. Red circles represent sensitivity and blue squares specificity. Solid curves are sigmoidal fit to data, with values for sensitivity and specificity maxima and minima constrained to 100% and 0%, respectively. SCV CI cutoffs at which sensitivity and specificity are equal and maximal are indicated by the vertical line. (A) Sensitivity and specificity for SCV CI cutoffs, as predictors for whether IVC CI is <20%, suggesting hypervolemia. The SCV CI cutoff of <22% corresponded to equivalent sensitivity/specificity of 72% Overall, these findings indicate that SCV max , IVC max , and collapsibility with respiration may be independently affected by both relative intravascular volume and significant (at least moderate) TR. The presence of clinically significant TR may bias the interpretation of venous maximum diameter and collapsibility with respiration resulting in overestimation of relative intravascular volume. 10 Because SCV CI and IVC CI were not significantly affected by the presence or absence of right-sided intracardiac devices or CKD stage, our data suggest that the use of SCV CI as a surrogate for IVC CI in the assessment of relative intravascular volume may be applicable to a general population of patients with ADHF that have right-sided intracardiac devices, and CKD Stages 2-4. SCV CI and IVC CI may also reflect RAP in patients with rate-controlled atrial fibrillation/flutter. 27 The limitations of our study are that this is a pilot study with a small but adequately powered sample size, and echocardiograms assessing the severity of TR were not obtained on the same day as SCV and IVC ultrasound.
In conclusion, SCV CI measured supine at 30-45°correlated well with paired IVC CI by ultrasound in non-ventilated patients hospitalized with ADHF and may provide an alternative to IVC CI for assessment of relative intravascular volume. SCV or IVC collapsibility with respiration may therefore be a useful adjunct to other available clinical information in assessing and managing ADHF. Moderate to severe TR decreases SCV CI and IVC CI and may result in overestimation of relative intravascular volume.