Acute lung injury (ALI) of septic and non-septic origin is a frequent cause of mortality in critically ill patients. During ALI, the inflammatory process in the lungs may increase the microvascular pressure and permeability, resulting in an accumulation of extravascular lung water (EVLW) and development of pulmonary oedema 1. However, it is difficult to estimate the amount of oedema fluid at the bedside. Clinical examination, chest radiography and blood gases have proven to be of limited value in quantifying pulmonary oedema 123. Several techniques to assess EVLW have therefore been developed.

Among the various methods for measurement of EVLW, thermo-dye dilution has been used most frequently 45678. In animal models of lung oedema, this method has been evaluated by comparison with postmortem gravimetry, which is supposed to be the 'gold standard' of EVLW measurements 789. In critically ill patients, fluid management guided by thermo-dye measured EVLW was associated with improved clinical outcome 10. Hence, EVLW has been suggested to play a role as an independent predictor of the prognosis and course of illness 6810. However, the thermo-dye dilution method is relatively time consuming, cumbersome and expensive. For these reasons, the method has not gained general acceptance 457.

Use of a technique based on injection of a single thermo-indicator that can be detected using an indwelling arterial catheter was an appealing concept. Recent experimental and clinical studies have shown that EVLW assessed by single thermodilution (ST) exhibits good reproducibility and close agreement with the thermo-dye double indicator technique 1112. The ST method is simpler to apply, less invasive and more cost effective; all of these factors make it more suitable for use at the bedside. However, to date, this new method has been sparsely evaluated against gravimetry 1314, and further validation is needed.

Thus, the aim of the present study was to evaluate the accuracy of the ST technique by comparing it with that of postmortem gravimetry (EVLW_G) in conscious sheep, in which ALI was induced either by lipopolysaccharide (LPS) or by oleic acid (OA). Both of these models of ALI are reproducible and have been extensively described 79111516.

All animals survived until the end of the experiments. At baseline no significant differences were found between groups, as shown in Figs 1 and 2, and Tables 1 and 2. In the sham-operated sheep, all variables remained unchanged throughout the study.

The present findings confirm that, in sheep, EVLW measured using the single transpulmonary thermodilution technique correlates closely with EVLW determined using postmortem gravimetry. However, EVLWI_ST overestimates EVLWI_G, with the degree of overestimation increasing with the severity of ALI.

A number of experimental and clinical studies focused on the potential role of EVLW as a guide to diagnosis and treatment of critically ill patients 3678910111213142526. During pulmonary oedema, accumulation of EVLW occurs before any changes take place in blood gases, chest radiogram and, ultimately, pressure variables. In addition, the latter variables are nonspecific diagnostic tools that are influenced by a variety of factors 2458. Thus, Boussat and coworkers 3 recently demonstrated that, in sepsis induced ALI, commonly used filling pressures such as PAOP and RAP are poor indicators of pulmonary oedema. Rather than those measures, they recommended direct measurement of EVLW. Consistent with this, we found that EVLW, in contrast to RAP, correlates with markers of lung injury in human septic shock 26. Victims of ALI, regardless of pathogenesis, have a significantly higher EVLW than do other patients 626. Hence, measurement of EVLW supports the diagnosis and may even improve clinical outcomes when used cautiously in combination with treatment protocols that are known to hasten the resolution of pulmonary oedema 1025.

Instrumented awake sheep represents a stable experimental model for measuring cardiopulmonary variables, as demonstrated in the sham-operated group in the present study as well as by other investigators 1527. The model can be used to assess different interventions during ALI.

Consistent with previous investigators 151727, we observed that infusion of LPS and OA caused pulmonary hypertension, increased EVLW and impaired gas exchange. Despite increments in PAP, PAOP and PVRI, both ITBV and GEDV remained constant whereas PVPI (an index of microvascular permeability, calculated as the ratio of EVLW to pulmonary blood volume) increased significantly. Thus, the haemodynamic responses to LPS and OA are not purely hydrostatic but may also manifest as noncardiogenic permeability pulmonary oedema 13151617182728.

In the present study lung oedema was significantly more severe in the OA group than in the LPS group, which is consistent with the findings of other investigators 29. In fact, OA causes acute haemorrhagic alveolitis, which may lead to acute endothelial and alveolar necrosis and a severe proteinaceous oedema 30. In contrast, the LPS-induced ALI is initiated by accumulation of granulocytes and lymphocytes in the pulmonary microcirculation that results in more moderate damage to endothelial cells and lung oedema 31.

Lung injury in the LPS group was accompanied by a hyperdynamic circulatory state, which was manifested by systemic vasodilation and increments in CI and DO₂I toward the end of the experiment. In contrast, in the OA group we observed cardiac depression and systemic vasoconstriction. This is consistent with previous investigations of LPS and OA 18273032. Thus, ovine models exhibit a scatter of cardiopulmonary changes from normal in the sham-operated group to mild or moderate ALI in endotoxaemic sheep and moderate to severe ALI in animals subjected to OA.

The significant correlation of EVLWI_ST and EVLWI_G observed in the present study is consistent with findings of Katzenelson and coworkers 13, who validated EVLWI_ST versus postmortem gravimetry in dogs 13. However, those investigators did not specifically assess the relationship between EVLWI_ST and EVLWI_G in sepsis-induced ALI. In addition, their study was performed in anaesthetized and mechanically ventilated animals; hence, further investigation of the correlation in a conscious state was required. Recently, ST has been evaluated against the thermo-dye dilution method in both experimental and clinical settings 1112. The studies revealed a close agreement between the techniques. Thus, we believe that injection of cold saline can provide valuable information about the EVLW content and the severity of pulmonary oedema.

During ALI, both ST and postmortem gravimetry demonstrated similar relative increases in EVLWI as compared with sham-operated animals. However, we noticed that ST overestimates the absolute values of EVLWI compared with the gravimetric technique – a discrepancy that increased with progression of pulmonary oedema. This finding could be accounted for by heat exchange of the thermal indicator with extravascular intrathoracic structures, such as the walls of the large vessels and the myocardium, and by recirculation of the indicator 8. In addition, the coefficients for calculation of EVLWI_ST and ITBV may vary with weight and age, as well as between animal species 11. Consequently, in the experimental setting EVLWI_ST requires a specific correction. In the present study we replaced the coefficient 1.25 used in humans in the ITBVI equation (i.e. ITBVI = 1.25 × GEDVI) with the recalculated 'ovine' coefficient 1.34 14, which is based on 426 measurements in 48 animals 171819.

In contrast to ST, the thermo-dye dilution technique runs the risk of underestimating EVLW in comparison with gravimetry 4. This underestimation increases during ALI caused by instillation of hydrochloric acid into the airways, and has been explained by redistribution of pulmonary blood flow away from the oedematous areas. The redistribution is thought to prevent indicator diffusion and consequently to prevent detection of oedema 7. In addition, detection of EVLW by thermo-dye dilution can be impaired by changes in CI as well as by positive end-expiratory pressure during mechanical ventilation 828.

Compared with other techniques for assessment of EVLW, ST may underestimate EVLW during pulmonary oedema due to intratracheal instillation of saline, although it is an accurate method in normal lungs 33. However, intratracheal instillation of saline can also be criticized because a proportion of the fluid is rapidly absorbed and obscured from detection 34.

Notably, the use of postmortem gravimetry as the reference method for evaluating pulmonary oedema also has limitations 2133. For example, the method only allows one measurement and is therefore of no use in following variations over time. The application of gravimetry is limited almost exclusively to experimental studies. The comparison of gravimetric measurement with results of other techniques for determination of EVLW can be influenced by the duration from death to removal of the lungs and by pathophysiological changes in the lungs after cardiac arrest. Thus, the gravimetric technique can underestimate the real value of EVLWI because of partial reabsorption of fluid before excision of the lungs.

The determination of EVLW by ST in sheep correlates closely with gravimetric measurements over a wide range of changes, and thus it may potentially be of benefit in quantifying lung oedema in critically ill patients. However, compared with postmortem gravimetry, single transpulmonary thermodilution overestimates the absolute values of EVLW. Thus, further studies are warranted to evaluate the accuracy of this method for managing ALI in humans.

• In sheep, extravascular lung water assessed by transpulmonary single thermodilution correlates closely with gravimetric measurements over a wide range of changes.

• Despite a moderate overestimation of the extravascular lung water content compared with post-mortem gravimetry, single thermodilution can be a useful tool for assessment of pulmonary oedema during ALI.

This study was supported by Helse Nord (Norway), project number 4001.721.132; departmental funds, the Department of Anesthesiology, University Hospital of North Norway; and Pulsion Medical Systems (Germany).

MYK participated in the design of study, performed statistical analysis, and drafted the manuscript. VVK participated in the design of study, performed statistical analysis, and prepared the figures. VVK and KW participated in the design of study. LJB participated in the design of study and provided coordination. All authors read and approved the final manuscript.

ALI = acute lung injury; CI = cardiac index; DO₂I = oxygen delivery index; EVLW = extravascular lung water; EVLWI = extravascular lung water index; GEDV = global end-diastolic volume; GEDVI = global end-diastolic volume index; ITBV = intrathoracic blood volume; ITBVI = intrathoracic blood volume index; LPS = lipopolysaccharide; OA = oleic acid; PAOP = pulmonary arterial occlusion pressure; PAP = pulmonary arterial pressure; PVPI = pulmonary vascular permeability index; PVRI = pulmonary vascular resistance index; Qs/Qt = venous admixture; RAP = right atrial pressure; ST = single thermodilution.