Diagnostic performance of pulmonary ultrasonography and a clinical score for the evaluation of hydration status in haemodialysis patients

BACKGROUND There is no feasible benchmark in daily routine to estimate the hydration status of haemodialysis patients, which is essential to their management. We performed a retrospective study in HD patients to assess the diagnostic performance of pulmonary ultrasound and clinical examination for the evaluation of fluid overload using transthoracic echocardiography (TTE) as a gold standard. METHODS 31 patients receiving chronic HD were included. Evaluation of hydration status was assessed weekly before haemodialysis sessions using clinical and Echo Comet Scores from pulmonary ultrasound and TTE (reference method). RESULTS Five patients had a TTE overload. Compared with TTE, the diagnostic performance of the clinical overload score has a sensitivity (Se) of 100%, a specificity (Sp) of 77%, a positive predictive value (PPV) of 50% and a negative predictive value (NPV) of 100% with a κ of 0.79. Only orthopnoea (P=0.008), jugular turgor (P=0.005) and hepatic-jugular reflux (P=0.008) were significantly associated with TTE overload diagnosis. The diagnostic performance of Echo Comet Score by pulmonary ultrasound has a Se of 80%, a Sp of 58%, a PPV of 26% and a NPV of 94%. Ten patients (32.3%) had an increase of extravascular pulmonary water without evidence of TTE or clinical overload. CONCLUSIONS Our clinical score has a convincing diagnostic performance compared to TTE and could be easily used in daily clinical routine to adjust dry weight. The evaluation of the overload using pulmonary ultrasound seems poorly correlated with the overload evaluated by TTE. The presence of extravascular pulmonary water undetected by clinical examination and TTE remains a parameter which requires further investigation.


Background
The definition of hydration status to determine optimal dry weight (DW) is essential for the management of haemodialysis patients. The DW concept in dialysis was conceived at the same time as dialysis in 1967 (1) and has evolved over time. DW is defined as the lowest weight tolerated by the patient at the end of the dialysis session at which there are minimal symptoms of hypovolemia or hypervolemia (2).
DW is an element of standard care for haemodialysis patients because inadequate DW is associated with increased cardiovascular morbidity and mortality (3)(4)(5)(6)(7).
Patients with end-stage renal disease (ESRD) requiring haemodialysis have a high risk of developing pulmonary congestion. The accumulation of pulmonary water can be infra-clinical and occurs gradually between dialysis sessions, especially in anuric patients (8) and is related in part to increased alveolar-capillary permeability (9).
Clinical examination is the classic tool for assessing hydration status at the patient's bedside. However, it can be faulted in finer evaluation. There is currently neither a validated clinical score to assess hydration status in haemodialysis patients nor a gold standard to define the DW. Various anthropometrical (10) or radiological tools (11)(12)(13)(14)(15) and biomarkers (16)(17)(18)(19) have been tested. Among them, transthoracic echocardiography (TTE) provides a more accurate assessment of blood volume than clinical examination(20) by studying left ventricular filling pressure (LVFP), systolic pulmonary arterial pressure (sPAP) and inferior vena cava (IVC) diameter at the same time, but it requires training and its long duration is poorly suited to routine use.
In lung ultrasound, B-lines are artefactual images resulting from close contact between (alveolar) air and water (clogged septa). Lung ultrasound detects pulmonary congestion with a sensitivity (Se) and a specificity (Sp) respectively, of 93 and 93% in intensive care patients (21). This technique has several advantages: it is fast (from 3 to 10 minutes) (12,14,22), non-irradiating and inexpensive. It can be performed with any ultrasound machine. The training is simple (about 2 hours) (13), with low inter-operator variability (23,24). It can be used in daily practice to detect fluid overload (FO) and has been validated in congestive heart failure and intensive care (25,26).
Recently, several studies have evaluated this technique in the haemodialysis population (27). The number of B-lines is correlated with an elevated LVFP by TTE, and the increase of total and lung water by bio-impedance spectroscopy (BIS) (11)(12)(13)(28)(29)(30). The number of B-lines decreases during a dialysis session and is correlated with weight loss and water loss in BIS (13,28). The relationship between the presence of asymptomatic pulmonary water on ultrasound, the DW and the occurrence of adverse events is yet to be clarified.
We conducted a retrospective study to compare the performance of lung ultrasound and clinical examination with TTE as the diagnostic gold standard to assess FO in haemodialysis patients.
The secondary objective was to determine which clinical signs best correlate with pulmonary and cardiac FO. months were included. All patients had at least three dialysis sessions per week. All patients gave their express consent. Procedures followed were in accordance with the Helsinki Declaration of 1975, as revised in 2000. Patients who missed more than one dialysis session in the previous month and patients with a history of pulmonary fibrosis or active lung infection were not included. The persistence of a residual diuresis was assessed for all patients and was defined by a diuresis volume superior to 500 mL per day.

Fluid overload assessment
The hydration status evaluation was performed during the mid-week dialysis session by a physician trained in clinical examination, cardiac and pulmonary ultrasound.
We have developed a clinical score to define the presence of FO using the following criteria: Major criteria -dyspnoea New York Heart Association (NYHA) > III, orthopnoea; Minor criteria -jugular turgor and hepatic-jugular reflux in half-sitting position, pulmonary crackles at auscultation, peripheral oedema (evaluated by searching an indentation after pressing the two inferior limbs (over the dorsum of the foot, behind and above the medial malleolus) and the sacral region) and predialysis high blood pressure. The association of two major or three minor criteria or the combination of one major and one minor criterion defined FO ( Table 1). This score was obtained at the beginning of the dialysis session.
Cardiac and pulmonary echocardiography was evaluated in the first 30 minutes of the dialysis session using an ultrasound machine (Philips® CX50 POC, Amsterdam, Netherlands), after the clinical examination. The TTE evaluated three parameters: 1) IVC diameter using the two-dimensional motion-mode method (M-mode), measured in sub-xiphoid view in the hepatic portion at non-forced end-expiratory and endinspiratory phases. IVC collapsibility index was calculated as follows: (maximum IVC diameter at expiration -minimum IVC diameter) / maximum IVC diameter *100. A collapsibility index greater than 40% is the threshold to define hypovolemia in spontaneous ventilation (31). 2) sPAP was evaluated by measuring the tricuspid regurgitation velocity peak plus the estimated right atrial pressure. The right atrial pressure was rated at 10 mmHg if the IVC diameter was greater than 2 cm and at 5 In the absence of a validated echocardiographic score to define overload in haemodialysis patients, we defined the echocardiographic FO as E/E' ratio > 13 or a combination of the following criteria: E/E' ratio between 8 and 13, IVC collapsibility <40% and sPAP> 35 mmHg. This definition was chosen using semiotic evidence bundles conventionally used in clinical practice and values described in the literature for ESRD patients (38)(39)(40)(41)(42). We chose this definition as the gold standard in our study because of the absence of a suitable benchmark to assess FO in ESRD patients, and because it has been shown that the E/E' ratio correlates well with the elevation of LVFP in cardiac catheterization in patients with ESRD (38). In addition, the thresholds appear to be similar to those in the non-haemodialysis population (41), with elevated values representing an independent risk factor for mortality (38).
Lung ultrasound assessed the number of anterior and lateral B-lines in a supine position. Lung water quantification was evaluated by the Echo Comet Score (ECS) using the 28-region technique described by Jambrik et al. (22). The sum of the Blines at each site led to a score of over 280 indicating the importance of extravascular pulmonary water. The presence of B-lines was considered "mild" (5 to 14 B-lines), "moderate" (15 to 29 B-lines) or "severe" (more than 30 B-lines) (39).
Clinical, anthropometric and demographic characteristics were collected at baseline. Intra-dialytic hypotension was defined as systolic blood pressure fall more than 20 mmHg, or more than 10 mmHg associated with signs of poor tolerance, according to Kidney Disease Outcomes Quality Initiative (KDOQI) recommendations (40).
Biological data were collected before the dialysis session. The relative changes in blood volume during the dialysis session were evaluated by the relative blood volume (RBV) monitor incorporated into the dialysis machine (Nikkiso® DBB05, Tokyo, Japan).

Statistical analysis
Categorical variables were tested by Chi-square test and expressed as counts and percentages. The quantitative values were tested by a Mann-Whitney test and expressed as median and interquartile (IQR) ranges (25th-75th percentiles) and calculation of correlation by Spearman test and linear regression. Inter-observer variability was measured by Cohen's kappa (κ) coefficient between 2 observers in 10 patients. All tests were non-parametric. A P<0.05 was considered significant. We express the diagnostic weight as likelihood ratio (LR) to describe the discriminatory power of clinical examination and lung ultrasound to define FO compared to the diagnostic gold standard (TTE). LRs were calculated using the Evidence-Based Medicine Calculator (©Knowledge Translation Program). Values greater than 1 increase the probability of disease. LRs less than 1 decrease the probability of disease. LRs of 2, 5, and 10 increase the probability of disease by about 15%, 30%, and 45%, respectively (in absolute terms). LRs of 0.5, 0.2, and 0.1 decrease probability by 15%, 30%, and 45%, respectively. We analysed clinical examination and lung ultrasound findings with tables comparing LRs of each different parameters to express the greatest diagnostic value (43).

General characteristics of the study population
Thirty-one patients were included between December 2016 and April 2017.
Characteristics of the study population are shown in Table 2. 83.9% of the patients were classified as hypertensive. Antihypertensive medications were prescribed to 67.8% of the study population. Most commonly prescribed were beta-blockers (32.3% of patients), followed by renin angiotensin system blockade (12.9% of patients), and calcium channel blockers (9.7% of patients  Figure 1). There was no significant difference between the TTE overload and no TTE overload groups for all other clinical signs. There was no significant difference in terms of the occurrence of intra-dialytic hypotension between the two TTE groups ( Table 3).

Clinical and TTE characteristics of the patients based on the presence of pulmonary water at lung ultrasound
There were not significantly more patients in overload according to the clinical score in patients with overload on the pulmonary ultrasound than in patients without pulmonary overload: 35.7 versus 29.4%, P=0.50. Ten patients (32.3%) had pulmonary water without clinical overload. Pulmonary water ultrasonography was significantly associated with the presence of crackles: 28.6% versus 0.0%, P=0.03.
There was no significant difference between the two lung ultrasound groups for all other clinical signs. There were not significantly more patients with TTE overload in patients with pulmonary ultrasound overload than in patients without lung overload: 5.8 versus 28.6%, P=0.11. Pulmonary water ultrasonography was significantly associated with a higher sPAP: 7.5 mmHg [5.0-11.5] versus 25.5 mmHg [5.0-39.7], P=0.012. There was no significant difference between the two lung ultrasound groups regarding other TTE data.

Discussion
In our study, it appears that lung ultrasonography data do not correlate with TTE data to assess FO in ESRD patients. It had a poor Sp and PPV, but a good NPV. FO clinical score appears to be a better tool for gauging TTE-assessed overload with greater Se, Sp, PPV, and NPV than lung ultrasound.
Clinical overload evaluated by the score proposed in this study appears well correlated with TTE overload data, in particular, orthopnoea, jugular turgor and hepatic-jugular reflux. Our score has the advantage of being simple and fast for a volume assessment in routine clinical practice, and with a good inter-observer reliability. Thus, it seems useful for detecting intravascular overload and would make it possible to avoid the realization of TTE to assess DW.
While TTE data (elevation of LVFP, sPAP, diameter and collapsibility of IVC) are markers of increased intravascular pressure, lung ultrasound seems rather to be a reflection of extravascular overload. It does not appear to be correlated with the echocardiographic data or clinical examination in our study. It was only correlated with crackles. Thus, it seems of interest for the detection of infra-clinical pulmonary FO, which is not identifiable using TTE. It could refine the accuracy of DW determination, particularly in patients for whom clinical evaluation of FO is difficult, since stating normohydration from only the intravascular fluid accumulation may lead to persistent fluid overload.
Because of the simplicity, the speed, and the excellent inter-observer reproducibility (42) of this examination, it appears a useful technique in which nephrologists should be trained.
Eleven of the 12 studies currently published on haemodialysis pulmonary ultrasound used ECS (27). Our results are consistent with those of the Lung Water by Ultrasound Guided Treatment in Hemodialysis Patients (LUST) study, which reported the low sensitivity of pulmonary crackles and peripheral oedema compared to lung ultrasonography for evaluation of lung water in haemodialysis patients, but a good specificity (29). Indeed, it has been shown that a decrease in ECS correlates with weight loss between the beginning and end of dialysis, while neither the diameter of the IVC (12,14) nor the E/A ratio varies (44). In our study we confirm the existence of a correlation between LFVP evaluated in TTE and ECS, which is consistent to other studies (23,45).
This suggests that the extravascular compartment would balance more slowly than the intravascular area with ultrafiltration. The study of Agricola et al. showed a linear correlation between ECS and extravascular pulmonary water determined in transpulmonary thermodilution (45). In our study, 15 out of 31 patients had overload on pulmonary ultrasound. However, the occurrence of a single episode of cardiac decompensation in our cohort suggests that in most cases this overload is well tolerated. The fact that the majority of the patients in the study had a residual diuresis probably limited the risk of cardiac decompensation.
Several questions remain unanswered. Our study shows 32.3% of clinically euvolemic patients with pulmonary water on ultrasound. There may be a risk of overtreating these patients. Ultrafiltration volumes that are too high lead to an exposure to risks, such as the occurrence of inter-dialytic hypotensions (46), fistula thrombosis (46,47), loss of residual diuresis (48,49) or decreased LVEF (50).
Asymptomatic lung congestion is probably also dependent on other variable factors such as vascular hyper-permeability (due to a possible endothelial dysfunction) (9), as albuminemia was not different in our study between patients with TTE overload and no TTE overload. Detection of asymptomatic lung water by ultrasound could allow for better control of the hydration status and avoids the occurrence of cardiac events. The current LUST randomized trial (ClinicalTrials.gov identifier No. NCT02310061) may answer this question by comparing the mortality and the risk of cardiovascular events in chronic haemodialysis patients according to management based on a daily clinical volume assessment or pulmonary ultrasound examination in everyday practice.
Ours is the first study comparing clinical signs with cardiac and pulmonary ultrasonography to detect FO at the chronic dialysis patient's bedside in stable ESRD patients. The physicians in charge of the patients were blinded to study results that could have influenced the change in DW. The fact that clinical evaluations were performed before ultrasound evaluations limits an eventual assessment bias, as the ultrasound findings appear to be more objective than the clinical findings.
Our work has several limitations. It is a relatively small sample drawn from a single dialysis centre. The clinical and TTE evaluation scores were chosen empirically, in the absence of clearly defined and validated scores in the haemodialysis population in the literature. We did not evaluated serum BNP in our study due to his poor sensitivity and specificity to assess fluid overload in dialysis patients (19). We did not evaluated BIS neither, as this technique was not available in our center and already compared to ECS in previous studies (28,30).

Conclusions
In conclusion, evaluation of overload using the ECS by pulmonary ultrasound is poorly correlated with the overload evaluated by TTE. The presence of extravascular pulmonary water undetected by clinical examination and TTE remains a parameter which requires further investigation from a diagnostic and prognostic point of view in haemodialysis patients. The clinical score proposed in this study has a satisfying diagnostic performance compared to TTE with good inter-observer reliability, and could be easily used in daily clinical routine to adjust DW.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and analysed during the current study are available from the corresponding author on reasonable request.

Competing Interests
The authors declare that they have no competing interests.

Funding
None declared.  Hepatic-jugular reflux -Predialysis blood pressure above 150/100 mmHg Clinical overload was defined by the association of 2 major criterions, or the association of 3 minor criterion the association of 1 major criterion and one minor criterion    Figure 1 Probability of fluid overload in echocardiography depending on clinical signs and lung ultraso