In recent years, there have been several investigations of the adult respiratory distress syndrome (ARDS) that are remarkable for the extensive measurements performed and for the new insights that they provide. [1,2] In this issue, Benzing et al. (page 1254) from the University of Freiburg, Germany, provide another such study. These authors measured indices of pulmonary oxygen exchange and vascular characteristics in 11 patients with severe ARDS, whose gas exchange was being supported by venovenous extracorporeal lung assistance. By adjusting the flow rate through the extracorporeal circuit, the mixed venous tension (PvO2) was systematically adjusted to 47, 54, 64, and 84 mmHg, whereas the carbon dioxide tension and cardiac output (Qt) remained unchanged and in the presence and absence of 15 ppm inhaled nitric oxide (NO). Addition of NO was observed to cause a greater reduction of perfusion pressure (PP) at low PvO2than at high PvO2, whereas reduction of shunt was almost constant (Table 1) at all PvO2. Identification of the mechanisms underlying such apparently dissociated responses is of special interest.
Ventilation was with 100% O2, and with the highest PvO2, the fractional shunt (fQs) was 0.674. Hypoxic pulmonary vasoconstriction (HPV) would be minimal during these conditions, and therefore throughout the study, 68% of the lung can be represented as one non-ventilated compartment in parallel with the remaining 32% that is ventilated. When Rt, Ra, and Rb are the total, nonventilated, and ventilated lung vascular resistances, respectively, then 1/Rt =(1/Ra + 1/Rb), and distribution of flow to the nonventilated (fQs) and ventilated (1 - fQs) compartments is given by [(1 - fQs)/fQs]= Ra/Rb. Table 1lists the essential observed and derived values. Benzing et al. explains these results intuitively with the assumptions that HPV acts only on the nonventilated compartment, and NO acts only on the ventilated compartment. These conditions can be simulated with a sophisticated computer model (V/Q - P/Q model ), which demonstrates three important principles illustrated in Figure 1. First (Figure 1(A)), even if there is no HPV present and no other cause of small artery constriction (SAC), so that fQs is constant, the PaO2increases dramatically as PvO2is increased. Second (Figure 1(B)), if HPV alone is present, then PP decreases with increasing PvO2, but NO has no effect. Conversely, if small pulmonary arteries are actively constricted but if HPV is absent, then NO reduces PP (and shunt), although by a constant amount independent of PvO2. Only when HPV and SAC are present (Figure 1(C)) is a PvO2-dependent response to NO revealed, although shunt also changes similarly. This computer simulation and consideration of the equations applicable to the simple model of two parallel compartments do not demonstrate dissociation of the pressure and shunt responses to NO, and therefore the assumptions of Benzing et al. are reexamined. Undoubtedly, HPV is inactive in the hyperoxic ventilated compartment,  but the pattern of changes shown in Table 1, for Ra and Rb when NO was administered, suggest that the action of NO is not confined to the ventilated compartment.
One difficulty with comparing Ra and Rb are the different sizes of the compartments. It is useful to normalize these resistances to what they would be if the whole lung had either the resistance, aTOTof the nonventilated compartment or the resistance, bTOT, of the ventilated compartment so Ra = aTOT/fa and Rb = bTOT/fb, where fa and fb are the fraction of lung in each compartment and are 0.68 and 0.32 for the data of Benzing et al. Thus, the total vascular resistance of each of the normalized parallel compartments (aTOTand bTOT) derived from the data of Benzing et al. can be reconstructed as three resistances in series consisting of HPV, SAC, and everything else (OTHer). Thus, aTOT= aHPV+ aSAC+ aOTHand bTOT= bHPV+ bSAC+ bOTH. Quantities can be assigned to each of these resistances with the following assumptions: bHPVis zero throughout; aOTH= bOTH; aTOTand bTOTfor the highest PvOsub 2 differ only by the minimal aHPVand bSACis zero with NO at the lowest PvO2. The estimated values are shown in Table 2and reveal as expected that aHPVdecreases systematically as PvO2increases, but more interestingly, the changes induced by the administration of NO differ in two important respects from the assumptions. First, although the small artery constriction of the ventilated lung (bSAC) decreases each time NO is administered, the change decreases as PvO2increases. Second, with administration of NO to the ventilated lung, the small artery constriction, aSAC, and the strength of HPV, aHPV, of the nonventilated lung also decrease. The changes again are reduced with increasing PvO2. Therefore, the inhibitory effects of NO are exerted on the ventilated and nonventilated compartments, and the efficacy of the NO action on either compartment is oxygen-dependent, as has been shown with other pulmonary vasoactive compounds. 
With these modifications, the V/Q-P/Q model is able to simulate the observed results of Benzing et al., and therefore the apparently dissociated changes in PP and shunt are accounted for quantitatively. PP is determined by PP = Qt/(1/Ra + 1/Rb). At the lowest PvO2, change of PP with NO is a result of reduction of resistance in both compartments, and the reduction in the change of PP with NO as PvOsub 2 increases is a result of the oxygen-dependent loss of efficacy of NO and the interaction with the decreased influence of HPV. Conversely, shunt is determined by fQs = 1/[(Ra/Rb)+ 1] so that the relatively constant change of shunt with NO at all PvO2results from the approximately proportionate change of the Ra/Rb ratio, which also is a result of the same influences and is presumably independent of the size of the nonventilated compartment. It is not surprising that NO acts on both compartments because in patients with ARDS, the nonventilated and ventilated lung regions are distributed contiguously throughout the lungs. However, this effect and the nature and extent of the oxygen dependence of NO action require confirmation and further study.
Bryan E. Marshall, M.D., F.R.C.A., F.R.C.P.
Carol Marshall, Ph.D.
Department of Anesthesia; Hospital of the University of Pennsylvania
781 Dulles; 3400 Spruce Street; Philadelphia, Pennsylvania 19096