Lung ultrasound: a ‘B-line’ to the prediction of decompensated heart failure

This editorial refers to ‘Detection and prognostic value of pulmonary congestion by lung ultrasound in ambulatory heart failure patients’†, by E. Platz et al. on page 1244. Despite advances in medical and device-based therapy for the management of ambulatory patients with heart failure (HF), the post-discharge mortality and readmission rates have continued to be unacceptably high,1,2 while chronic mortality has actually increased recently.3 Central to frequent hospitalizations and high mortality among patients with HF is cardiogenic pulmonary oedema, characterized by the accumulation of fluid in the lung's interstitial and alveolar spaces at least partially due to elevated cardiac filling pressures. Early accurate diagnosis and subsequent treatment of pulmonary congestion may prevent hospitalizations. Furthermore, reliable quantitative assessment of the pulmonary interstitial fluid may also aid the provider in determining the most appropriate discharge date, preventing hospital readmission. Lung ultrasound (LUS) has emerged as a readily available, efficient, semi-quantitative method for assessing pulmonary congestion. Previous work has shown that LUS is effective as a diagnostic tool, with a higher sensitivity and specificity than the clinical exam and chest X-ray.4 Platz and colleagues now report the prognostic significance of LUS in ambulatory patients with New York Heart Association (NYHA) class II–IV HF and a previous HF hospitalization in the past 12 months. Two-second ultrasound clips were recorded in eight LUS zones by trained investigators using a pocket ultrasound device. The number of B-lines visualized is an estimate of the degree of pulmonary congestion. B-lines are sound artefacts created from an acoustic impedance mismatch that occurs when a sound beam crosses two interfaces (e.g. fluid between lung and chest wall). The highest number of B-lines visualized in any single intercostal space was recorded for each zone. The sum of B-lines in all eight zones was used for primary analysis. Ambulatory outpatients who had adequate LUS data in all eight zones (n = 185/195 evaluated or 95%) were followed for 6 months. The baseline characteristics, stratified by B-line tertiles, revealed that patients with the highest number of B-lines had higher risk features (higher NYHA class, worse renal function, higher NT-proBNP, and lower sodium and haematocrit). Patients with the highest B-line number were also more likely to have elevated jugular venous pulsations, crackles on pulmonary auscultation, and peripheral oedema. In another report, higher B-lines were associated with echocardiographic evidence of elevated filling pressures (E/e'), lower 6-min walk distance, and clinical congestion on a standard clinical congestion scale.6 Thus, B-lines may be a reliable maker for lung congestion in HF patients, and may serve as a risk stratification tool when other indicators are difficult to interpret. Next, the investigators found that patients with the highest tertile of B-line quantity had a four-fold higher risk of the primary outcome [adjusted hazard ratio (HR) 4.08, 95% confidence interval (CI) 1.95–8.54, P < 0.001] compared with those patients in the first tertile. This finding was primarily driven by the higher number of hospitalizations in patients in the highest tertile. Patients in the lowest tertile spent a significantly lower number of days alive and out of the hospital (125 days vs. 165 days, adjusted P < 0.001) compared with those in the third tertile. Finally, using the incremental discrimination improvement (IDI) and the area under the receiver operating characteristic (ROC) curve (AUC), the authors suggested that LUS provides incremental prognostic value compared with physical exam findings (crackles on auscultation; congestion score). Unfortunately, only a subset of patients had NT-proBNP drawn at a time close to their clinic visits, and so the incremental value of LUS over biomarkers was not assessed. The work of Platz and colleagues should be applauded, as it provides an important step forward for the implementation of LUS in the clinical evaluation of HF patients. When compared with most of the established clinical tools for detecting decompensation, LUS has performed similarly.6 In a recent prospective study in which emergency physicians were asked to categorize the diagnosis for patients presenting with acute dyspnoea, LUS had a significantly higher accuracy (sensitivity 97%, specificity 97%) in differentiating acute decompensated HF from non-cardiac causes than the initial clinical work-up, chest X-ray alone, and natriuretic peptides.4 In the present study, the authors found that only 19% of patients in the highest tertile had crackles on auscultation, suggesting that LUS may be able to identify subclinical congestion and improve risk stratification. However, the lack of a true gold standard for pulmonary congestion and acute decompensated HF is an important limitation that prevents the understanding of the true accuracy of LUS. While LUS certainly has the benefits of being more accessible and faster than echocardiography, there is no standard technique established to examine the lungs with ultrasound. Both the 28-scanning-site technique and the 8-region technique have been recommended as useful in the evaluation of pulmonary interstitial oedema.7 In the study of Platz et al., eight LUS zones (four on each hemithorax) were scanned, and the sum of B-lines in all eight zones was used for analysis. Previous studies have included a varied number of chest sites scanned, various patient positions (supine, orthopnic, sitting), and have used a varied number of B-lines as cut-off values for risk stratification. Furthermore, while all studies claim that the LUS exam can be performed within 10 min, and often within <5 min, all investigations have utilized experienced sonographers. Real-world experience with emergency department staff, primary care providers, and cardiology clinic providers may not achieve such rapid exam times, calling into question the incremental utility of LUS in a patient with elevated neck veins, peripheral oedema, and an elevated NT-proBNP. Perhaps the most exciting application of LUS may be in the prognostic value of B-lines at discharge to predict rehospitalization. Two small prospective studies have recently shown that residual pulmonary congestion before discharge, as assessed by LUS, predicts rehospitalization and all-cause death in patients admitted with HF. 8,9 In this context, LUS may be a very useful tool in identifying patients that need continued hospitalization for therapy optimization vs. those that are safe for discharge. Continued investigations utilizing LUS must be pursued in the future. The present work advances the field of LUS in HF by correlating the number of B-lines to HF mortality and future hospitalizations. Early recognition of patients with impending decompensation by LUS when other variables are equivocal may aid the clinician in risk stratification (Figure 1). However, previous work has found that non-invasive biomarkers as well as implantable monitoring devices that detect persistent or developing congestion are very good at predicting risk in HF patients, but are equivocal in guiding therapy and improving outcomes.10,11 Thus, the most clinically relevant unanswered question remains: are B-lines on LUS a therapeutic target? In a different patient population, this question is being addressed in the Lung Water by Ultrasound Guided Treatment in Hemodialysis Patients (LUST; clinicaltrials.gov NCT02310061) study. It is imperative to test the hypothesis of whether LUS-guided therapy of HF patients could improve outcomes. Therefore, additional work is necessary to demonstrate that routine LUS assessment is a non-invasive method that not only predicts, but can also prevent HF morbidity and mortality. A potential conceptual approach for the integration of lung ultrasound (LUS) in the risk stratification and management of patients with heart failure is presented. This algorithm is dependent upon the future demonstration that the information provided by LUS can meaningfully impact clinical outcomes. Conflict of interest: none declared.