INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS INDICATOR DILUTION MODEL FOR... RESULTS DISCUSSION REFERENCES

ALVEOLAR HYPOXIA RESULTS in constriction of small pulmonary arteries. Direct observations on dog pulmonary vessels have revealed narrowing of arteries in the 30- to 800-µm diameter range (2, 3, 24, 25). With panalveolar hypoxia, such as that resulting from decreased inspired PO₂ or hypoventilation, the larger pulmonary arteries tend to be distended by the elevation in pulmonary arterial pressure resulting from the constriction of the smaller vessels (3). Pulmonary veins in the 30- to 1,000-µm diameter range also have been observed to constrict but much less so than the arteries over most of that range (3, 24). Similar observations have been made in cat (28, 32), rabbit (17), and ferret (5, 31) lungs.

The impact of alveolar hypoxia on the pulmonary capillary bed, either as an indirect result of the constriction of muscular arteries and veins or as the result of some mechanism for directly affecting capillaries (12, 20, 26, 27, 33, 37), is not as clear. One approach to evaluating the effects of hypoxia on the pulmonary capillary bed has been to measure the diffusing capacity for CO, DL_CO, as an index of capillary blood volume. The results have been interpreted as indicating a decrease in pulmonary capillary volume with hypoxia (18, 36, 37). However, because both capillary volume and PO₂ affect DL_CO, there are some uncertainties associated with this interpretation. Capen and Wagner (4) attempted to avoid this problem by measuring the DL_CO during hypoxia with and without infusion of the pulmonary vasodilator prostaglandin E₁. Thus they could reverse the hypoxic vasoconstriction without changing the PO₂. They concluded that since vasodilation in the presence of hypoxia did not increase DL_CO, hypoxia alone must have increased capillary volume. This, along with their direct observation of hypoxia-induced recruitment of apical subpleural capillaries (34), led them to conclude that alveolar hypoxia increases capillary volume by recruitment of vessels that are not perfused during the normoxic condition (4). In a study of cat lungs, no significant change in the morphometrically determined capillary blood volume was detected in response to alveolar hypoxia (33). Thus there are apparently conflicting conclusions as to the effect of acute hypoxia on the pulmonary capillary volume.

Another related observation is that acute alveolar hypoxia has been consistently found to decrease total vascular volume in lungs of various species (1, 10, 12). This brings up the question of whether the constriction of the small arteries and veins can displace a sufficient volume to account for the decrease in total volume in the face of increased large arterial volume (2, 3, 25) and possibly in capillary volume (4). In studies wherein we injected boluses of contrast medium to visualize the arteries and veins in perfused dog lung lobes (3), we observed that when we placed a region of interest (ROI) over an area of lung that had no vessels larger than 50 µm, the area under the X-ray absorbance vs. time curve following the injection of a bolus of contrast medium was smaller during hypoxia than during normoxia. This area is proportional to the vascular volume within the ROI (7). Another consideration in the total accounting of effects of hypoxia on the distribution of the pulmonary blood volume is that the available morphometric data indicate that the volume in arteries and veins smaller than 50 µm makes up less than 5% of the volume in all vessels smaller than 50 µm (16, 22). This would appear to implicate a capillary contribution to the hypoxia-induced decrease in total volume. Therefore, we decided to follow up on our X-ray observations to determine whether they would shed light on the question of the effect of hypoxia on the pulmonary vascular volume distribution in dog lungs. We also compared the results obtained with hypoxia with those obtained during infusion of serotonin [5-hydroxytryptamine (5-HT)], another pulmonary vasoconstrictor stimulus but with a different distribution of sites of action (3, 12), and with passive manipulation of capillary volume due to pressure change.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS INDICATOR DILUTION MODEL FOR... RESULTS DISCUSSION REFERENCES

X-ray images of isolated, perfused dog lung lobes [N = 25, 18.8 ± 4.3 (SD) kg body wt, 38.6 ± 10.9 g lung lobe wet wt] were acquired as described previously (3, 6, 14). One hour prior to anesthesia, each dog received an oral dose of ~1 g of aspirin to ensure that a consistent hypoxic response would be obtained in the isolated lung lobe preparation when the PO₂ in the ventilating gas mixture was reduced (3, 12). The dog was then anesthetized (30 mg/kg pentobarbital sodium) and heparinized (1,250 IU/kg). After exsanguination, via a carotid arterial catheter, the chest was opened, and the left lower lobar artery and vein were cannulated. The lobe was excised, and the arterial and venous cannulas were attached to the perfusion system that was primed with the dog's own blood [average hematocrit of 29.4 ± 4.1 (SD)]. The perfusion system included a roller pump that pumped the blood from a reservoir into the lobar artery at a flow of 6.47 ± 1.40 (SD) ml/s. The blood then drained back through the lobar vein into the reservoir.

The lobar bronchus was attached to the ventilation system, and the lobe was ventilated between measurements with a gas mixture containing ~15% O₂-5.6% CO₂-79.4% N₂ during control conditions. This resulted in a PO₂ of 116 ± 7 (SD) Torr, PCO₂ of 42 ± 3 (SD) Torr, and pH of 7.31 ± 0.07 (SD) in the recirculating blood. The end expiratory transpulmonary pressure was 3.5 ± 0.5 (SD) Torr.

The inflow tubing included an injection loop that allowed the introduction of a bolus of radiopaque contrast medium (3-4 ml of 61% iopamidol, Isovue 300) into the lobar arterial inflow, without changing the pressure or flow (3). The lobe was positioned between the X-ray source and detector of the imaging system [either Nicolet NXR-200 (35-µm focal spot) with an X-ray-sensitive video camera (13 lobes) or Fein Focus FXE-100.20 (3-µm focal spot) with an image intensifier/charge-coupled device camera (12 lobes)]. Just prior to bolus injection, the ventilation was stopped at end expiration, and the lobe was held at the end expiratory pressure until the bolus had cleared the field of view. Video images were recorded at 30 frames/s using a Super-VHS VCR as the bolus of contrast medium passed through the lobar vasculature.

The field of view at the settings used was ~12 cm in diameter at low magnification and 1 cm in diameter at high magnification. The locations of the high-magnification fields of view were chosen such that, considering all of the lung lobes combined, they were distributed evenly over the lobar silhouette. The original observations were made at high magnification. Low-magnification imaging was added to obtain a more complete accounting of the lobar transit times. Thus high-magnification images were obtained on all 25 lobes, and low-magnification images were obtained on a subset of 15 lobes.

Hypoxia was produced in 18 of the lobes by ventilating them with a gas mixture containing ~5% O₂-6% CO₂-89% N₂, which reduced PO₂ in the recirculating blood to an average of 39 ± 10 (SD) Torr and produced an increase in perfusion pressure of about 9 mmHg. 5-HT was infused into the lobar artery of 12 lobes. The infusion rate was adjusted to produce an increase in perfusion pressure of about 6.3 mmHg. This resulted in an average infusion rate of 38 ± 19 (SD) µg/min. In seven of the lobes, to produce a passive change in vascular volume, i.e., without vasoconstriction, the venous pressure was raised.

Time sequences of images acquired from the same regions during control and at least one of the three experimental conditions (hypoxia, 5-HT, and high pressure) were analyzed to determine total lobar and/or microvascular volume changes as follows. ROIs were positioned over the inlet lobar artery, A, and outlet lobar vein, V, on the low-magnification images to acquire lobar inlet and outlet curves. On high-magnification images, ROIs were positioned over a small feeding artery (500 µm diameter) and the nearby microvasculature free of any visible arteries or veins (i.e., free of vessels larger than ~50 µm). Up to 20 microvascular ROIs were positioned within each field of view. X-ray intensity vs. time curves were acquired from each ROI as the bolus passed through the vasculature within the field of view. The acquired intensity curves were then converted to absorbance curves, Abs(t), by the log transformation

(1)

where I(t) is the measured X-ray intensity at time t, I(0) is the average baseline intensity measured before the arrival of the bolus, and I_min is the minimum measured intensity corresponding to maximum X-ray absorbance (6, 7). Thus, at low magnification, Eq. 1 was used to generate the inlet, c_A(t), and outlet, c_V(t), absorbance (proportional to concentration) curves for the whole lung lobe from curves obtained from the A and V ROIs.

From high-magnification images, time-absorbance curves were obtained from the small artery ROI, a(t), and an adjacent microvascular ROI, m(t). Both a(t) and m(t) were time averaged over three consecutive time points, corresponding to a 0.1-s time interval. Absorbance due to the small artery only, c_in(t), excluding the surrounding microvasculature, was obtained by subtracting m(t) from a(t) as previously described (6-8). This subtraction is based on the assumption that m(t) measures the background microvascular absorption in the path of the X-ray beam within the lobe, which also contributes to the measured a(t) curve. The resulting background-corrected arterial absorbance curve was used as a locally representative microvascular inlet concentration curve, c_in(t), whereas the microvascular ROI absorbance curve represents the local microvascular residue curve, m(t).

	INDICATOR DILUTION MODEL FOR IMAGE ABSORBANCE CURVES

TOP ABSTRACT INTRODUCTION METHODS INDICATOR DILUTION MODEL FOR... RESULTS DISCUSSION REFERENCES

In an imaging indicator-dilution experiment in which a bolus of contrast material is injected, images are acquired as the bolus passes through the field of view with the contrast medium serving as the vascular indicator. Absorbance curves acquired from the arterial inlet, c_A(t), and venous outlet, c_V(t), of the whole lung lobe can be used to estimate the total mean transit time (µ_lobe, first moment) and variance of transit times (_lobe², second moment about the mean) through the lobe (8, 39). That is

(2)

and

(3)

where T denotes the time point at which the integrand (i.e., the respective absorbance curve) crosses zero after having reached its peak value. Lobar vascular volume is then computed as the product of lobar mean transit time (µ_lobe) and lobar flow.

Furthermore, regional analysis may be performed by acquiring the small inlet artery absorbance curves, c_in(t), which are proportional to the concentration of contrast medium passing through the small artery at time t, and microvascular ROI absorbance curves, m(t), which are proportional to the mass of contrast medium in the microvascular ROI at time t. In other words, m(t) is given by the mass-balance equation

(4)

where c_in(t) and c_out(t) are the ROI inlet and outlet absorbance curves, respectively, and F is the constant flow rate through the ROI (6, 7, 39). By writing c_out(t) = c_in(t) * h(t), where h(t) represents the frequency distribution of transit times through the ROI, and the asterisk denotes temporal convolution, Eq. 4 becomes

(5)

If we now define

(6)

Eq. 5 becomes

(7)

where the flow-weighted impulse residue function, R_F(t), represents the m(t) curve that would have been measured from the ROI had the concentration curve at the ROI inlet been an impulse. Because convolution is area preserving

(8)

Thus, from Eq. 8, a change in the ratio of the mass, m(t), area to the concentration, c_in(t), area represents a proportional change in ROI volume.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS INDICATOR DILUTION MODEL FOR... RESULTS DISCUSSION REFERENCES

Table 1 presents the hemodynamic data obtained. The lobar arterial pressure, P_art, was significantly higher than control during either hypoxia, 5-HT infusion, or high pressure.

Figure 1 is an example of selected images from the time sequence acquired during bolus passage through the vasculature of one lung lobe during control, hypoxia, and 5-HT infusion. To highlight subtle gray-level differences, these example images have had the preinjection (t = 0) image subtracted and are displayed in pseudo-color. Figure 1, left, shows the arterial tree filled with contrast medium obtained at the peak of c_A(t) (see Fig. 2). Figure 1, middle, shows an image obtained when c_A(t) on its downslope was equal to c_V(t) on its upslope, in which case the contrast medium resides mainly within the background microvasculature having mostly left the arterial tree and not yet filling the venous tree. Figure 1, right, obtained at the peak of c_V(t), shows the contrast medium draining through the venous tree. The most striking observations are that during hypoxia the image in Fig. 1, hypoxia middle, is less intense than the control or 5-HT images. In addition, at the peak of c_V(t), the veins have more substantially emptied. The 5-HT image intensity in Fig. 1, 5-HT middle, is more like that of the control, although the appearance is more heterogeneous.

View larger version (153K): [in this window] [in a new window]   Fig. 1.   Images of an isolated dog lung lobe acquired during the passage of a bolus of contrast medium (flow rate = 4.7 ml/s, body wt = 11 kg) through the vasculature during control (top row), hypoxia (middle row), and serotonin (5-hydroxytryptamine; 5-HT) infusion (bottom row). In each case, the prebolus background image was subtracted prior to display. Arterial tree (left) and venous tree (right) images were acquired at time of maximum absorbance within the inlet lobar artery (A, left) or outlet lobar vein (V, right) region of interest (ROI). ROIs A and V also indicate locations from which absorbance curves in Fig. 2 were obtained. Microvascular images (middle column) were acquired at the crossover time at which the lobar inlet artery and outlet vein absorbance curves shown in Fig. 2 were equal.

View larger version (16K): [in this window] [in a new window]   Fig. 2.   Lobar inlet arterial (cA) and outlet venous (cV) absorbance curves acquired from ROIs A and V, respectively, in Fig. 1, during control and hypoxia (A), 5-HT (B), and high pressure (C).

Figure 2 shows absorbance curves acquired from the inlet lobar artery, c_A(t), and outlet lobar vein, c_V(t), of the lung lobe imaged in Fig. 1 during control, hypoxia (Fig. 2A), 5-HT (Fig. 2B), and high-pressure (Fig. 2C) conditions. In this case, the increase from control in lobar arterial pressure was 10.7, 11.0, and 7.0 mmHg for hypoxia, 5-HT, and high pressure, respectively. With hypoxia and 5-HT, c_V(t) is shifted to the left, relative to the control c_V(t), reflecting the decreased lobar mean transit time and, since flow was held constant, decreased lobar vascular volume. With high pressure, the transit time and volume were increased, as indicated by the right-shifted c_V(t).

The mean values of total lung lobe mean transit time, µ_lobe, and variance, _lobe², were determined using Eqs. 1 and 2 applied to inlet/outlet curves such as those shown in Fig. 2 from all conditions. The resulting lung lobe relative dispersion (RD_lobe = /µ_lobe) and vascular volume (µ_lobe × flow) were also computed and are given in Table 2. Hypoxia and 5-HT infusion both resulted in significantly decreased lobar transit time and hence decreased lobar vascular volume. Furthermore, the relative dispersion increased significantly from control in both cases. Increasing arterial pressure resulted in a significant increase in lobar mean transit time and vascular volume but a decrease in relative dispersion.

Figure 3 illustrates the bolus passage through a high-magnification lung lobe region during control (Fig. 3, top) and hypoxia (Fig. 3, bottom). Analogous to Fig. 1, the arterial (Fig. 3, left) and microvascular (Fig. 3, right) image times were chosen to correspond to maximum absorbance within the small artery (a) and microvascular (m) ROIs. Again, as in Fig. 1, the microvascular panel of Fig. 3 (right) is less intense during hypoxia than during control. One can also observe the hypoxia-induced decrease in diameter of the smaller arterial vessels.

View larger version (165K): [in this window] [in a new window]   Fig. 3.   Background-subtracted, high-magnification images of bolus passage (flow rate = 4.9 ml/s, body wt = 20.9 kg) through the same lung region during control (top) and hypoxia (bottom). Artery (left) and microvascular (right) images were acquired at the time of maximum absorbance within the arterial ROI (a, 420-µm diameter, left) or microvascular ROI (m, containing only vessels of control diameter <50 µm, right).

Microvascular volume changes were studied by analyzing absorbance curves obtained from ROIs over the same location during the control and three experimental conditions. Figure 4 illustrates c_in(t) and m(t) curves acquired from a and m ROIs, respectively, of images such as those shown in Fig. 3. The control and hypoxia (Fig. 4A) or 5-HT (Fig. 4B) curves were obtained from the same ROIs, whereas curves obtained from another region during both control and high-pressure conditions are shown in Fig. 4C. With hypoxia, unlike 5-HT infusion, the area under m(t) is greatly reduced, relative to the control m(t), suggesting decreased microvascular ROI volume. With high pressure, the microvascular volume was increased, as indicated by the increased area under m(t).

View larger version (16K): [in this window] [in a new window]   Fig. 4.   Small artery (cin) and microvascular residue (m) curves obtained from the same ROIs during control and hypoxia (A), control and 5-HT (B), and control and high pressure (C).

The volume formula of Eq. 8 was used to estimate the regional microvascular volume parameter for each microvascular ROI from the measured inlet and residue curves. The fractional change in microvascular volume within each ROI, (volume_exp  volume_con)/volume_con, where volume_con is the volume at control and volume_exp is the volume during either hypoxia, 5-HT, or high pressure, was computed for each condition and is plotted in Fig. 5. The fractional change in lobar vascular volume with each experimental condition is also shown in Fig. 5.

View larger version (24K): [in this window] [in a new window]   Fig. 5.   Fractional change from control in lobar vascular (open bars) and microvascular (hatched bars) volume during hypoxia, 5-HT, and high pressure. Asterisks indicate change is significantly different from 0, with *P < 0.05, **P < 0.01, or ***P < 0.001. Lobar vascular volume change is significantly different from microvascular volume change (P < 0.05). Numbers at the bottom indicate the number of reported lungs (N), microvascular ROIs (n), and average number of ROIs per lung (avg ROIs/lung) for each condition.

Several observations can be made from Fig. 5. First, as also indicated in Table 2, total lobar vascular volume decreased significantly with both hypoxia (P < 0.001) and 5-HT (P < 0.01), and the magnitude of that decrease was not statistically different between these two stimuli. When subjected to high vascular pressure, total lobar vascular volume increased significantly (P < 0.05) from control.

Second, as also shown in Fig. 5, hypoxia caused a significant decrease in microvascular volume (P < 0.001). Furthermore, the fractional decrease in microvascular volume was significantly larger than the fractional decrease in lobar vascular volume (P < 0.05). The microvascular response to 5-HT was quite variable, and the mean was not significantly changed from control. Elevated vascular pressure resulted in increased microvascular volume (P < 0.001), with approximately the same fractional increase as in the total lobar vascular volume.

To evaluate the uniformity of the microvascular response to hypoxia throughout the lobe, the location of the high-magnification imaging field of view was chosen differently for each lobe, such that cumulatively the entire lobar image was covered. The distances from the ROIs to the hilum ranged from ~2 to 9 cm. Furthermore, in one experiment, 67 small arterial and microvascular ROI pairs, with distance from the hilum ranging from 1.8 to 9.2 cm and average distance of 6.7 ± 2.1 (SD) cm, were examined in a low-magnification image. Although in this low-magnification analysis, the ROIs probably contained somewhat larger vessels than in the high-magnification studies, all of the ROI volumes except one decreased with hypoxia. The average fractional change in ROI volume of 0.21 ± 0.01 (SE) was consistent with the hypoxic microvascular volume change obtained with high-magnification ROIs (Fig. 5).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS INDICATOR DILUTION MODEL FOR... RESULTS DISCUSSION REFERENCES

These results are consistent with the observation made in several previous studies that hypoxic vasoconstriction decreases total pulmonary blood volume (1, 10, 12, 30, 38). They also provide additional information as to how the decrease is distributed. Hypoxia clearly decreased microvascular volume. The relationship between microvascular and capillary volume is perhaps somewhat less clear because capillaries and the smallest arteries and veins are not individually observable. However, given that the extant anatomic data attribute such a small volume to the small arteries and veins (16, 22), a decrease in capillary volume appears consistent with the results. It is not clear how such a decrease might come about. In the present study, the flow was held constant. Thus there is no reason to believe that the capillary pressure would fall in response to an increase in upstream resistance, and an increase in downstream resistance would increase capillary pressure. Closure of some small arteries resulting in cessation of flow in some capillaries could decrease capillary volume accessible to the contrast medium. Derecruitment as the result of hypoxic vasoconstriction may be consistent with the concept of arterial smooth muscle tone acting as a critical closing pressure (29). However, closure of small arteries in response to hypoxia has apparently not been seen on direct observation of subpleural pulmonary arteries (21). Small arteries disappear from the images such as in Fig. 3, but it is not clear whether some actually close or simply fall below the limits of resolution.

The capillaries are commonly thought to be devoid of contractile machinery for adjusting lumen dimensions. However, Kapanci et al. (26, 27) and Weibel (35) noted that there are cells within the lungs that appear capable of mediating pulmonary capillary response. Hillier et al. (24) and Groh et al. (17) have observed narrowing of small pulmonary arteries in response to hypoxia in the size range commonly categorized as nonmuscular, further emphasizing the plausibility of an active "nonmuscular" capillary response.

Direct observations on subpleural capillaries have indicated an increase in subpleural capillary recruitment and presumably in capillary volume in response to hypoxia (4). Questions have been raised as to whether the subpleural vessel behavior reflects that of the interior capillaries (19). As indicated in the introduction, Capen and Wagner (4) addressed this question by measuring diffusing capacity, which is a measure of the volume of the interior capillaries, with their subpleural capillary perfusion index. Thus the reasons for the apparent differences between the present study and subpleural capillary observations are not clear. One obvious difference in the experimental preparations is that the subpleural capillary observations were made on intact dogs, whereas the present experiments were carried out on isolated lungs. In the intact dogs, the cardiac output increased during hypoxia, whereas the flow was held constant in the present experiments. This may make some contribution to the apparent difference, and whether the overall response in the intact animal and in the isolated lung are the same is probably not a key point. Instead the observation that hypoxia can produce a decrease in pulmonary capillary volume even under the specific conditions of these experiments indicates an aspect of the complexity of the microvascular response to hypoxia that is not explained.

At this point, the possibility that capillary narrowing or derecruitment occurs during hypoxia has no clear implication with regard to the contribution of the capillaries to the hypoxia-induced change in total pulmonary vascular resistance. Knowledge of the capillary contribution to the total resistance during normoxia and additional insight into the geometry of the capillary bed would be required to address this question. This knowledge is still incomplete (23).

5-HT infusion produced a much more heterogeneous microvascular response than hypoxia. Direct observation of the images revealed a blotchy image during the microvascular passage of the bolus as exemplified by Fig. 1 (bottom). Although it is possible that this reflects a heterogeneity in some aspect of the machinery (e.g., receptors) involved in the 5-HT response, it seems likely that at least part of the heterogeneity reflects the difficulty in providing an even flow-proportional distribution of the small volume of 5-HT-containing infusate at the arterial inlet (i.e., in obtaining adequate mixing). This is as opposed to decreasing the inspired PO₂, which provides an evenly distributed stimulus. Thus, although infused 5-HT and hypoxia produce different longitudinally distributed patterns of pulmonary arterial constriction (3), and the overall microvascular response was more heterogeneous with 5-HT, the local microvascular response may be similar for the two stimuli.

The data obtained when the vascular pressure was changed might be thought of as a positive control demonstrating that a passively induced capillary volume change was reflected in m(t). Under the assumption that capillary pressure is about midway between arterial and venous pressure, the fractional change in microvascular volume of about 3% per mmHg is consistent with previous measurements of capillary compliance of the dog lung obtained using other methods (14, 22).

This is apparently the first application of X-ray angiography and ROI residue detection to hypoxia-induced changes in pulmonary microvascular volume. The method quantifies the fractional changes in volume, where, as indicated above, the size of the microvascular vessels is defined by exclusion. The actual c_in(t) to a given ROI is not measurable, because a given microvascular ROI may have multiple inlets and because, although vessels as small as 50 µm can be visually distinguished during bolus passage, the contrast in such small vessels does not provide sufficient signal for actually quantifying c_in(t). Therefore, R_F(t) itself is not available. However, only the area of R_F(t), which is equal to the area of m(t), is required for the volume measurement. We normalized the m(t) areas to the c_in(t) areas obtained from small arteries (~500 µm) within the given field of view. However, consistent with indicator dilution theory, the c_in(t) areas are virtually invariant under the various study conditions as exemplified in Fig. 4. That is, as long as the bolus volumes and the X-ray technique factors (beam-intensifier-object geometry, voltage, and current) were constant, the normalization was not actually necessary to quantify the volume changes. The fact that the arterial tree, microvascular bed, and venous trees can be separately visualized as time progresses during bolus passage as shown in Fig. 1, along with the fact that most of the bolus dispersion takes place in the microvasculature (8), reflects the large fraction (50-60%; Refs. 8 and 14) of the total vascular volume contained in the microvascular bed.

The decrease in total vascular volume with hypoxia includes the volume accessible to vascular indicators that trace only the perfused volume, such as in the present study (11, 13, 38), and the total vascular volume detected by decreased lung weight (11, 12, 30). With constant flow perfusion, the volume of the larger pulmonary arteries increases as the result of the increase in pressure due to narrowing of the small vessels (3, 11) wherein the volume obviously decreases. From the present study, it appears that a decrease in capillary volume may also contribute to the total volume decrease. Further evaluation of the mechanisms responsible would appear to be necessary to provide a more complete understanding of the pulmonary microvascular response to hypoxia.