Extravascular lung water assessed by transpulmonary single thermodilution and postmortem gravimetry in sheep

The study was approved by the Norwegian Experimental Animal Board and conducted in compliance with the European Convention on Animal Care. Eighteen yearling sheep weighing 27.5 ± 0.4 kg were instrumented, as a modification to previously described techniques [16‐19], by inserting introducers into the left external jugular vein and common carotid artery. After 1–4 days of recovery, sheep were placed in an experimental pen. A thermodilution catheter (131HF7; Edwards Life Sciences, Irvine, CA, USA) was introduced into the pulmonary artery and a 4-Fr thermistor-tipped catheter (PV2014L16; Pulsion Medical Systems, Munich, Germany) into the carotid artery. The catheters were connected to pressure transducers (Transpac^®III [Abbott, North Chicago, IL, USA] and PV8115 [Pulsion Medical Systems], respectively).

Mean pulmonary arterial pressure (PAP), pulmonary arterial occlusion pressure (PAOP) and right atrial pressure (RAP) were displayed on a 565A Patient Data Monitor (Kone, Espoo, Finland) and recorded on a Gould Polygraph (Gould Instruments, Cleveland, OH, USA). Heart rate, mean systemic arterial pressure, cardiac index (CI), systemic vascular resistance index, extravascular lung water index (EVLWI) assessed using the single thermodilution technique (EVLWI_ST), pulmonary vascular permeability index (PVPI), global end-diastolic volume (GEDV) index (GEDVI), intrathoracic blood volume (ITBV) index (ITBVI) and blood temperature were determined at 1-hour intervals using a PiCCOplus monitor (Pulsion Medical Systems). Every value reported here is the mean of three consecutive measurements, each consisting of a 10 ml bolus of ice-cold 5% dextrose injected into the right atrium randomly during the respiratory cycle.

To estimate EVLW we used the following formula [12]: EVLW_ST (ml) = ITTV - ITBV (where ITTV is the intrathoracic thermal volume). During clinical application of ST by means of the PiCCO monitor, ITBV is calculated as 1.25 × GEDV, the coefficient 1.25 being derived from critically ill patients [12]. However, in our previous investigations in sheep [17‐19], in which ITBV was measured directly using the thermal-dye dilution technique, we found the coefficient to be 1.34 [14]. Thus, in the present study we used the corrected values of ITBVI, EVLWI_ST and PVPI, based on the following equation: ITBVI = 1.34 × GEDVI.

Blood samples were drawn from the systemic arterial (a) and pulmonary arterial (v) lines and analyzed every two hours for blood gases and haemoglobin (Rapid 860; Chiron Diagnostics Corporation, East Walpole, MA, USA). The pulmonary vascular resistance index (PVRI), venous admixture (Qs/Qt), oxygen delivery index (DO₂I) and oxygen consumption index were calculated as described previously [16, 19, 20].

After establishing a stable baseline at time 0 hours, awake and spontaneously breathing sheep were randomly assigned to three experimental groups: a sham operated group (n = 4); a LPS group (n = 7), receiving an intravenous infusion of Escherichia coli O26:B6 LPS (Sigma Chemical, St. Louis, MO, USA) at 15 ng/kg per min for 6 hours; and an OA group (n = 7), in which sheep were subjected to an intravenous infusion of OA (Sigma Chemical) 0.06 ml/kg mixed with the animal's blood. The duration of the infusion of OA was 30 min.

During the experiment, all animals received a continuous infusion (5 ml/kg per hour) of Ringer's lactate, aiming to maintain intravascular volume at baseline levels. After the last measurements, at 2 hours in the OA group and at 6 hours in the sham-operated and the LPS groups, the sheep were anaesthetized and killed with a lethal dose of potassium chloride. Then, postmortem EVLWI (EVLWI_G) was determined by gravimetry, as previously described [21‐24].

For each continuous variable, normality was checked using the Kholmogorov-Smirnov test. Data are expressed as mean ± standard error of the mean, and assessed by analysis of variance followed by Scheffe's test or test of contrasts, when appropriate. To evaluate the relationship between EVLWI_ST and EVLWI_G, we used linear regression and Bland-Altman analysis. P < 0.05 was considered statistically significant.

Figure 1 and Table 1 show that LPS and OA induced marked increments in PAP and PVRI, peaking at 1 hour and subsequently decreasing gradually to values significantly above the respective baselines and the corresponding values in the sham-operated group. PAOP and RAP also rose in both the LPS and the OA groups (P < 0.05; data not shown). In parallel, LPS increased EVLWI_ST transiently by 20–35% (P < 0.05; Fig. 1). After OA administration, EVLWI_ST rose to a maximum of 84% above baseline (P < 0.01). At the end of the experiment, EVLWI_ST in the OA group had increased by 6.4 ml/kg and 9.3 ml/kg relative to the LPS and the sham-operated groups, corresponding to increments of 54% and 104%, respectively (P < 0.05). PVPI increased by 40% after LPS administration and by 90% after OA (P < 0.05; Fig. 1). GEDVI and ITBVI varied within 10–15% of baseline with no intergroup differences. As shown in Table 1, LPS caused tachycardia and a rise in CI accompanied by a slight increase in mean arterial pressure whereas systemic vascular resistance index decreased (P < 0.05). In contrast, in the OA group CI declined and systemic vascular resistance index increased relative to baseline (P < 0.05).

LPS caused significant increments in mixed venous oxygen saturation, DO₂I and Qs/Qt (Fig. 2). OA decreased both arterial and venous oxygenation and reduced DO₂I (P < 0.05). Oxygen consumption index did not change significantly (not shown). LPS caused a transient reduction in arterial carbon dioxide tension and a rise in pH (P < 0.05; Table 2). After OA, pH decreased (P = 0.04). The haemoglobin concentration as well as the body temperature rose only in the LPS group (P < 0.05).

As shown in Fig. 3, the regression analysis between EVLW_ST and postmortem EVLW_G yielded the following relation: EVLWI_ST = 1.30 × EVLW_G + 2.32 (n = 18, r = 0.85, P < 0.0001). Notably, the mean EVLWI_ST at the end of experiment was higher than EVLWI_G: 13.6 ± 1.1 ml/kg versus 8.7 ± 0.7 ml/kg (P = 0.0005). Ranges of EVLWI_ST and EVLWI_G values were 7.5–21.0 ml/kg and 4.9–14.5 ml/kg. According to the Bland-Altman analysis, the mean difference between EVLWI_ST and EVLWI_G was 4.91 ml/kg, with upper and lower limits of agreement (± 2 standard deviations) of +9.99 ml/kg and -0.17 ml/kg, respectively (Fig. 4). The difference between methods increased with increasing values of mean EVLWI (n = 18, r = 0.64; P = 0.005); the regression line equation was as follows: EVLWI_ST - EVLWI_G = 0.89 × ([EVLWI_ST + EVLWI_G]/2) + 6.82.

As shown in Fig. 5, EVLWI_G in the OA group increased by 4.7 ml/kg and 5.6 ml/kg relative to the LPS and the sham-operated groups, amounting to increments by 65% and 90%, respectively (P = 0.001).