INTRODUCTION: Ultrasound is an emerging method for assessing lung congestion but is still
seldom used. Lung congestion is an important risk of cardiac events and
death in end-stage renal disease (ESRD) patients on hemodialysis (HD).
OBJECTIVE: We investigated possible variables associated with lung congestion among diabetics with ESRD on HD, using chest ultrasound to detect extracellular lung water.
METHODS: We studied 73 patients with diabetes as the primary cause of ESRD, undergoing regular HD. Lung congestion was assessed by counting the number of B lines detected by chest ultrasound. Hydration status was assessed by bioimpedance analysis and cardiac function by echocardiography. The collapse index of the inferior vena cava (IVC) was measured by ultrasonography. All patients were classified according to NYHA score. Correlations of the number of B lines with continuous variables and comparisons regarding the number of B lines according to categorical variables were performed. Multivariate linear regression was used to test the variables as independent predictors of the number of B lines.
RESULTS: None of the variables related to hydration status and cardiac function were associated with the number of B lines. In the multivariate analysis, only the IVC collapse index (b = 45.038; p < 0.001) and NYHA classes (b = 13.995; p = 0.006) were independent predictors of the number of B lines.
CONCLUSION: Clinical evaluation based on NYHA score and measurement of the collapsed IVC index were found to be more reliable than bioimpedance analysis to predict lung congestion.
Keywords: extracellular fluid; kidney failure, chronic; pulmonary edema; ultrasonography.
INTRODUÇÃO: A ultrassonografia é um método emergente e ainda raramente utilizado na
avaliação da congestão pulmonar. A congestão pulmonar é um importante fator
de risco para eventos cardíacos e óbito entre pacientes com doença renal
terminal (DRT) em hemodiálise (HD).
OBJETIVO: Foram investigadas as possíveis variáveis associadas a congestão pulmonar em indivíduos diabéticos com DRT em HD, utilizando a ultrassonografia torácica para detectar água extracelular nos pulmões.
MÉTODOS: Foram estudados 73 pacientes com diabetes como causa primária de DRT submetidos a HD regular. A congestão pulmonar foi avaliada pela contagem do número de linhas B detectadas por ultrassonografia torácica. O estado de hidratação foi avaliado por análise de bioimpedância e a função cardíaca por ecocardiografia. O índice de colabamento da veia cava inferior (VCI) foi medido por ultrassonografia. Todos os pacientes foram classificados segundo a escore da NYHA. Foram analisadas as correlações entre o número de linhas B e variáveis contínuas e as comparações entre o número de linhas B em relação às variáveis categóricas. Regressão linear multivariada foi utilizada para testar as variáveis enquanto preditores independentes do número de linhas B.
RESULTADOS: Nenhuma das variáveis relacionadas a estado de hidratação e função cardíaca apresentou associação com o número de linhas B. Na análise multivariada, apenas o índice de colabamento da VCI (b = 45,038; p < 0,001) e as classes da NYHA (b = 13,995; p = 0,006) foram preditores independentes do número de linhas B.
CONCLUSÃO: A avaliação clínica baseada na classificação da NYHA e na medição do índice de colabamento da VCI foram mais confiáveis do que a análise de bioimpedância para predizer congestão pulmonar.
Palavras-chave: líquido extracelular; falência renal crônica; edema pulmonar; ultrassonografia.
|Citation: Santos PR, Lima Neto JA, Carneiro RAA, Soares AITD, Oliveira WR, Figueiredo JO, et al. Variables associated with lung congestion as assessed by chest ultrasound in diabetics undergoing hemodialysis. Braz. J. Nephrol. (J. Bras. Nefrol.) 39(4):406. doi:10.5935/0101-2800.20170073|
|Received: March 20 2017; Accepted: May 23 2017|
Chronic volume overload and left ventricular disorders are hallmarks of end-stage renal disease (ESRD). These two disorders lead to high prevalence of lung congestion among ESRD patients undergoing hemodialysis (HD). It is estimated that 60% of patients on maintenance HD present lung congestion.1 Besides hypervolemia and heart failure, high level of inflammation in ESRD patients contributes to lung congestion due to microvascular lung disease, provoking capillary leakage.2,3
Lung congestion is a strong prognostic marker of cardiac events and death among ESRD patients submitted to HD. HD patients with severe congestion have a 4.2-fold higher risk of death and a 3.2-fold higher risk of cardiac events, like myocardial infarction, angina, heart failure and arrhythmia.1
Clinical evaluation and bioimpedance analysis are daily tools in dialysis centers to estimate overhydration among HD patients in order to control lung congestion. However, extracellular volemia as assessed by bioimpedance is very weakly associated with lung water.4 Most importantly, extracellular and lung water were compared regarding their predictive power for adverse events. The conclusion is that lung water, and not extracellular water, is by far the most important predictor.4
Thus, how can lung water be detected? Pulmonary capillary wedge pressure is the most reliable method to estimate extravascular lung water, but this method of assessment is highly invasive. Recently, chest ultrasound has emerged as a safe, inexpensive and reliable method to measure lung congestion.5 Regardless of its simplicity, chest ultrasound is not yet incorporated in daily clinical evaluations. In daily practice, lung congestion is typically first evaluated by asking patients about symptoms, like dry cough, shortness of breath and dyspnea.
However, most patients even with moderate to severe lung congestion are asymptomatic.1 Secondly, lung water is estimated by assessing volume overload by bioimpedance. But, as stated above, total extracellular water may be not correlated with lung water. Thus, ultra-filtration prescription guided by estimation of overhydration may not control lung congestion.
In addition, the complications of targeting euvolemia in conventional HD are well known: episodes of hypotension, loss of residual renal function and myocardial fibrosis by repetitive episodes of myocardial ischemia.6,7 In summary, ultra-filtration based on overhydration may not control lung congestion and also poses well-known risks to HD patients.
Some authors argue in favor of some "permissive volume overload" in order to avoid the risks of targeting euvolemia.8 We think this "permissive volume overload" would be the amount of extracellular water with no lung congestion. However, the point is how to know whether or not lung congestion is present. Since chest ultrasound is still not widely used in dialysis centers, we thought it would be valuable to rely on reliable predictors of lung congestion among clinical and laboratory evaluations routinely performed among HD patients.
Diabetic patients on HD form a special group with high rate of mortality and cardiovascular disorders. This group of patients poses more challenges than other groups concerning the control of lung congestion. Thus, we conducted the present study aiming to find variables associated with lung congestion among diabetics with ESRD on maintenance HD.
We conducted a cross-sectional study with a sample selected among 305 ESRD patients on HD in June 2016 from the only two dialysis centers in an area of 34,560 km2 (37.3 inhabitants/km2) in the northern region of Ceará state, northeast Brazil. We included patients with diabetes as the primary cause of ESRD, older than 18 years and undergoing regular HD for at least three months.
The criteria for exclusion were extremity amputation (precluding bioimpedance analysis) and clinical instability with hospitalization. All of them were undergoing conventional HD (three sessions of four hours per week) with polysulfone dialyzers (maximum number of reuses = 12). Written informed consent was obtained from all participants, and the study was approved by the ethics committee of Vale Acaraú University, with which the hospital is associated.
Evaluation of lung congestion
Lung congestion was evaluated by counting the number of B lines by using chest ultrasound. B lines were recognized as a hyperechogenic, coherent bundle with narrow basis spreading from the transducer to the further border of the screen, as described by Bedetti et al.9 B lines are the ultrasonographic expression of interlobular pulmonary septa thickened by edema.
Chest ultrasound was performed by two observers from our research group of medical students trained by an experienced radiologist. If the difference in the number of B lines was under 10% between the two observers, the mean of the two values was considered. In case of differences greater than 10%, a new chest ultrasound was performed by a third observer. There was only one case of difference greater than 10% between the two observers.
Chest ultrasound was performed shortly before the bioimpedance procedure and before the first HD session of the week (on Monday for patients with HD scheduled for Mondays, Wednesdays and Fridays, and on Tuesday for patients with HD scheduled for Tuesdays, Thursdays and Saturdays). Patients were examined in the supine position. The right and left thoracic regions were examined.
The observers identified and quantified the B lines by positioning the 3.0 MHz probe in 28 positions: from the second to fifth intercostal spaces of the right hemithorax; from the second to fourth intercostal spaces of the left hemithorax; and in each intercostal space in four positions: parasternal, midclavear, anterior and middle axillary lines.
Hydration status assessment
Patients' hydration status was evaluated using bioimpedance analysis employing the spectroscopy technique, performed with a Body Composition Monitor® (Fresenius Medical Care, Bad Homburg, Germany). Patients underwent bioimpedance analysis just before the first HD session after the weekend, in other words, after the longest interval between dialysis sessions, and after performing chest ultrasound.
The bioimpedance procedure was conducted according to the manufacturer's manual by the nurses of the dialysis centers, all of them trained to use the Body Composition Monitor®. If any erroneous measurements were detected by the BCM on the basis of a measurement quality indicator, the respective measurement was repeated by the nurse.
Three variables were assessed: 1-extracellular water (ECW) in liters = amount of water in the body which is not inside cells (interstitial water plus the plasma water plus the transcellular water); 2-fluid overload (FO) in liters = the excess fluid stored almost exclusively in the extracellular volume of a patient, therefore part of the ECW; and 3-relative fluid overload (rFO), calculated in percentage as follows: FO/ECW x 100. Hypervolemia was classified as rFO > 15%.
Echocardiography was carried out by a cardiologist trained in echocardiography who assessed the following variables, according to the recommendations of the American Society of Echocardiography:10 left atrium volume (mm), diastolic thickness of the interventricular septum (mm), posterior wall diastolic thickness (mm), left ventricular diastolic diameter (mm), ventricular systolic diameter (mm), and ejection fraction of the left ventricle (%). Echocardiography was carried out during the same week in which chest ultrasound and bioimpedance were conducted.
Diameter of inferior vena cava
The same operator of echocardiography conducted the sonography procedure to measure the diameter (cm) of the inferior vena cava (IVC) by subcostal route. He measured this diameter after forced expiration and during inspiration. The IVC collapse index was calculated as follows: diameter in cm after forced expiration divided by diameter in cm during inspiration.
Classification by the new york heart association (NYHA) score
Patients were classified based on NYHA score by a doctor unaware of the results obtained by chest ultrasound, echocardiography, and sonography of vena cava. The NYHA scoring system stratifies patients into four classes.11 Class I is the one with the least symptoms related to heart failure and IV corresponds to the most intense symptoms. This score is widely used in clinical practice and has already been validated in a population of ESRD patients.12
Demographic and clinical data
Demographic data, duration of diabetes since diagnosis, length of time on dialysis, type of vascular access, blood pressure, interdialysis weight gain and volume of diuresis per 24 hours were obtained from the two dialysis centers' medical records. Volume of diuresis was recorded as it appeared in the medical records, stratified into four classes: zero diuresis (anuria)/24 h, up to 500 ml/24 h, from 500 to 1000 ml/24 h, and more than 1000 ml/24 h.
Classification of economic class was according to criteria of the form issued by the Brazilian Association of Research Institutes.13 This validated instrument is used in marketing surveys and population censuses and grades economic class into five subgroups: A (best status) through E (worst status). Besides income level, its criteria include educational level of the head of household and ownership of household appliances. Body mass index (BMI) was calculated as Kg/m2. Laboratory tests for serum creatinine, hemoglobin, albumin, calcium and phosphorus were performed. The dialysis dose delivered was evaluated using a second-generation Kt/V equation according to Daugirdas.14
The Shapiro test was used to assess normality of the distribution of continuous variables. Continuous variables with normal distribution are expressed as mean ± standard deviation, while those without normal distribution are expressed as median, minimum and maximum values. Pearson and Spearman tests were used to assess correlation between the number of B lines and continuous variables, respectively with and without normal distribution.
Comparisons regarding the number of B lines according categorical variables were performed by the Mann-Whitney test (between two groups) and Kruskal Wallis test (between more than two groups). Multivariate linear regression was used to test the continuous variables (which correlated with the number of B lines) and categorical variables (which differed regarding the number of B lines) as independent predictors of the number of B lines (dependent variable). Statistical significance was considered to be a p-value < 0.05. All the statistical analyses were performed using the SPSS version 22.0 program package.
Among the 305 ESRD patients undergoing HD, there were 87 with diabetes as the primary cause of ESRD. We excluded one patient due to age less than 18 years, four with less than three months of maintenance HD, six patients presenting extremity amputation, two clinically instable and hospitalized patients, and one who refused to participate. Therefore, we studied 73 patients.
Demographic and clinical data of the sample are shown in Table 1. Hydration status and echocardiography parameters are shown in Table 2. The only case of discordance due to the number of B lines observed by the examiners (58 versus 65 B lines) was decided as 65 B lines by a third observer. The number of B lines, as assessed by chest ultrasound, were positively correlated with left ventricular systolic diameter (r = 0.293; p = 0.030), left atrial diameter (r = 0.289; p = 0.036), septal thickness (r = 0.312; p = 0.021) and posterior wall thickness (r = 0.303; p = 0.025) as assessed by echocardiography, and with the IVC collapse index (r = 0.355; p = 0.008) as assessed by ultrasonography (Table 3).
|Gender, n (%)||45 (62.5)|
|Age, mean ± DP||60.6 ± 15.8|
|Economic class, n (%)|
|Duration of diabetes (years), median [min-max]||15 [1,5 - 47]|
|Time on dialysis (months), median [min-max]||23 [3 - 122]|
|Type of vascular access, n (%)|
|Native fistula||54 (75.0)|
|Double lumen catheter||18 (25.0)|
|Interdialysis weight gain (kg), mean ± DP||3,4 ± 1,7|
|Diuresis per 24 hours, n (%)|
|< 500 ml||46 (63.9)|
|500-1000 ml||13 (18.0)|
|1000 ml||4 (5.6)|
|Systolic blood pressure (mmHg), mean ± DP||162.1 ± 27.7|
|Diastolic blood pressure (mmHg), mean ± DP||77.3 ± 15.3|
|Body mass index (kg/m2), median [min-max]||26.1 [17.7-47.0]|
|NYHA class, n (%)|
|Glycemia pre-dialysis (mg/dL), median [min-max]||190 [80 - 390]|
|Creatinine (mg/dL), median [min-max]||5 [2.4 - 12.6]|
|Hemoglobin (g/dL), mean ± DP||8.3 ± 1.4|
|Albumin (g/dL), median [min-max]||4.1 [2.0 - 4.9]|
|Calcium (mg/dL), mean ± DP||8.7 ± 0.9|
|Phosphorus (mg/dL), mean ± DP||4.7 ± 1.4|
|Calcium-phosphorus product (mg2/dL2), mean ± DP||42.4 ± 14.3|
|Kt/V index, mean ± DP||1.8 ± 0.6|
|Extracellular water (L), mean ± DP||16.2 ± 3.0|
|Fluid overload (L), median [min-max]||2.2 [-7.5-5.6]|
|Relative fluid overload (%), mean ± DP||12.5 ± 10.0|
|Hypervolemia, n (%)|
|Inferior vena cava diameter (cm), mean ± DP||1.6 ± 0.3|
|Inferior vena cava collapse index||0.4 [0.07-1.0]|
|Left ventricular ejection fraction (%), median [min-max]||59.3 [33.6-66.0]|
|Left atrial diameter (mm), median [min-max]||37 [20.0-69.0]|
|Septal thickness (mm), median [min-max]||12 [9-16]|
|Posterior wall thickness (mm), median [min-max]||12 [9-16]|
|Left ventricular diastolic diameter (mm), mean ± DP||51.0 ± 5.4|
|Left ventricular systolic diameter (mm), median [min-max]||32.0 [19.0-49.0]|
|Variable||r (correlation coefficient)||p|
|Duration of diabetes||0.107||0.433|
|Time on Dialysis||0.036||0.791|
|Interdialysis weight gain||-0.106||0.470|
|Systolic blood pressure||0.139||0.306|
|Diastolic blood pressure||0.028||0.837|
|Inferior vena cava diameter||0.223||0.102|
|Inferior vena cava collapse index||0.355||0.008|
|Left atrial diameter||0.289||0.036|
|Posterior wall thickness||0.303||0.025|
|Left ventricular diastolic diameter||0.203||0.137|
|Left ventricular systolic diameter||0.293||0.030|
In the comparison regarding the number of B lines according the categorical variables, we found a difference regarding the number of B lines according to NYHA classes: medians of 10 in class I, 19 in class II, 30 in class III and 65.5 in class IV (p = 0.042) (Table 4). In the multivariate analysis, only the IVC collapse index (b = 45.038; p < 0.001) and NYHA classes (b = 13.995; p = 0.006) were independent predictors of the number of B lines (Table 5).
|Variable||Number of B lines (median)||p|
|< 500 ml||14|
|> 1000 ml||19|
|Type of vascular access|
|Variable||b (regression coefficient)||p|
|Inferior vena cava colapse index||45.038||< 0.001|
|Left atrial diameter||0.275||0.493|
|Posterior wall thickness||1.438||0.851|
|Left ventricular systolic diameter||0.899||0.204|
As others,4,15 we showed the lack of association between lung congestion and hydration status, as assessed by bioimpedance. This result leads to practical implications, since extracellular water assessed by bioimpedance is usually the guide for the ultra-filtration prescription of HD patients. Thus, the control of volemia based on extracellular water does not guarantee the control of lung water.
However, the benefit of controlling volemia is to preclude heart burden and mainly lung congestion. Lung water is the main marker of cardiovascular morbidity and mortality, not the total extracellular water. Moreover, to target dry weight based on overhydration imposes risks of episodes of hypotension, most of them asymptomatic. Repetitive episodes of hypotension due to excessive ultra-filtration, especially among patients who present large interdialysis weight gain, can lead to myocardial stunning, which can cause myocardial remodeling and fibrosis.6,7
Also, we did not find correlation of lung water with cardiac function. None of the differences found in univariate analysis regarding the parameters of echocardiography were significant in the multivariate analysis. Lack of correlation between cardiac function and lung congestion may be due to sample characteristics.
Comparing the studies of Mallamaci et al.16 and Siriopol et al., 4 we observe that cardiac function correlates with lung water only in the former. We suppose that the divergence related to the correlation between cardiac function, and lung congestion is partially explained by sample characteristics. Our sample, like the sample studied by Siriopol et al.,4 presented relatively good cardiac function and a narrow range of values of ejection fraction: 59.3% (33.6% to 66%) in ours and 61.5% ± 7.7% in the other study.4
In both studies, there was no correlation between cardiac function and extracellular lung water. On the other hand, in the study of Mallamaci et al.,16 they found a correlation of cardiac function with lung water, probably because their sample presented a wider range of values of ejection fraction: from 15% to 70%; and worse cardiac function among patients with lung congestion: ejection fraction of 49%.
Thus, we hypothesize that in samples with a better systolic function, lung water can be dissociated from extracellular lung water. The good cardiac function of our sample can be explained by a well-known survivor selection: diabetics with poorer cardiovascular condition with no adequate medical follow-up in underdeveloped areas often die from cardiovascular complications before starting HD.
Traditional predictors of fluid overload were not validated as variables associated with lung congestion in our study, like lower body mass index,17,18 less residual urine output,18 higher systolic blood pressure.18,19 In our study, the NYHA result (a very simple scoring based on clinical evaluation) and IVC collapse index (a well-known marker of hypervolemia) were associated with lung congestion.
We firmly believe that in a short time, routine evaluation of lung congestion by chest ultrasound will be incorporated in dialysis centers and will widely replace bioimpedance as the main guide for ultra-filtration prescription. Meanwhile, and according to our result, how can NYHA and IVC collapse index be useful? We suggest NYHA be used by doctors and nurses in a systematic way and assessed weekly.
Since measurement of IVC is not performed in routine echocardiography of HD patients, we propose that information on IVC diameter and its collapse index be incorporated in echocardiography records. Better than this would be the regular evaluation of IVC by dialysis center teams themselves, as a way to help define patients' dry weight. Finally, none of these parameters are substitutes for periodic anamnesis and physical examination.
Although chest ultrasound is a very practical method to assess lung congestion, neither follow-up of HD patients by chest ultrasound nor interventions based on its results have been shown to modify patients' outcomes in the literature. Our experience in this study was that the learning curve of performing chest ultrasound for this purpose by non-specialists is very fast. However, the method is totally dependent on the operator. Thus, the counting of B lines depends on the operator's skill.
We are aware of several limitations of our study. First, the study presents a cross-sectional design and most of the variables are dynamic. However, due to the exclusion of patients with less than three months of HD and those hospitalized, we studied only stable patients. Stable patients do not have big changes of cardiac function and of the pattern of interdialysis weight gain in the short term. Second, as stated in the introduction, inflammation probably acts as a mediator of pulmonary capillary leakage, predisposing the appearance of extracellular lung water.
Thus, it would have been better to have had data on inflammation markers, like C-reactive protein and proinflammatory cytokines. Also, cardiac biomarkers (D-dimer, troponin T, natriuretic peptide) would make the panel of variables more complete. We had albumin in our study, which is a marker of malnutrition-inflammation syndrome. However, low levels of albumin only appear in advanced stages of this syndrome. Finally, despite being a sub-group of patients with high cardiovascular risk, our sample presented a relatively good cardiac function. Therefore, our results cannot be extrapolated to more typical samples of diabetics with more compromised cardiac function.
Clinical evaluations based on NYHA score and on the IVC collapse index assessed by ultrasonography are associated with lung congestion. Routine use of these variables can help to establish a safe ultra-filtration goal able to control and avoid lung congestion without major cardiovascular risks. In our opinion, these indirect evaluations using predictors of lung congestion will soon be replaced by the incorporation of chest ultrasound to assess lung water directly and routinely.