Determination of extravascular lung water by dilution methods. I. Theoretical considerations

Extravascular lung water measured from the different dilution of water and intravascular tracers, injected upstream and sampled downstream to pulmonary circulation, underestimates actual extravascular water, Qwt, found postmortem in canine lungs. We investigated whether in the dilution method unwarranted assumptions are made on eliminating recirculation by double extrapolation (DE). We cleared dilution curves of recirculation by injecting tracers in superior vena cava and sampling them simultaneously from pulmonary artery aand aorta. Thus, for each tracer we obtained a pair of input-output curves (I-O) that, using deconvolution techniques, yielded the frequency function of water transit time in the extravascular lung space, wt(t). In 7 dogs and in 17 measurements in man, in control conditions and with pulmonary edema, wt(t) showed an early peak followed by a marked knee and a long tail. While the area of wt(t) was 0.99±0.08 in dogs and 0.98±0.06 in man showing practically complete recovery of water tracer, normalized variance of wt(t), β, was 1.83±0.62 in dogs and 2.53±0.85 in man, indicating large dispersion of wt(t). The mean of wt(t) estimated from DE was 58.2±20.5% of that obtained by I-O in all the 24 determinations in dog and man. This underestimation correlated (r=-0.67, P<0.001) with β. Thus, the shape of wt(t) appears to be a cause of DE underestimation of Qwt. Accurate measurements of Qwt can be obtained by dilution method provided recirculation is truly eliminated.