INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

WHEN PULMONARY BLOOD FLOW is pulsatile, vascular resistance falls (9, 16) and gas exchange improves (7). Both of these observations could be explained if pulsatile flow caused capillary recruitment, for recruitment of capillaries would increase the cross-sectional area of the microvascular bed and the surface area available for gas exchange. It is not known whether pulsatile perfusion itself will lead to capillary recruitment. We hypothesize that if mean arterial and left atrial pressures are held constant during the switch from steady to pulsatile flow, the systolic pressure rise will open capillaries (hypothesis 1) and that, once open, the capillaries will not close during diastole (hypothesis 2). We also predict that the increased systolic pressure will redistribute pulmonary blood flow from the dependent lung to the upper lung (hypothesis 3). We tested these hypotheses by perfusing the canine left lower lung lobe with autologous blood and changing the perfusion system between a steady-flow pump and a pulsatile-flow pump. The venous outflow height and pump flow rates were adjusted to maintain constant mean arterial and left atrial pressure as the pumps were switched. Perfusion of the subpleural microcirculation was recorded by videomicroscopy. Capillary recruitment and capillary transit times were measured under each condition. The distribution of pulmonary blood flow was determined under each flow condition by injection of fluorescent microspheres and, later, analysis of the air-dried lung to measure the microsphere distribution.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Animal preparation. The experiments were approved by the Animal Care Committee of Indiana University School of Medicine. Healthy adult male mongrel dogs (19-23 kg, n = 7) were anesthetized with pentobarbital sodium (30-40 mg/kg iv), intubated, and mechanically ventilated with room air via a constant-volume respirator (model 607D, Harvard Apparatus). After injection of heparin (1,000 U/kg), each animal was rapidly exsanguinated through a cannula (3 mm ID) placed in the left common carotid artery. A thoracotomy was performed in the left fifth intercostal space, and the left main pulmonary artery was cannulated with a Teflon fluorinated ethylene polypropylene cannula (6 mm ID) via the right ventricular outflow tract. This arterial cannula was secured with a ligature placed around the left pulmonary artery and a second ligature placed around the main pulmonary artery and aorta to prevent loss of blood through the aorta from the left ventricle. A polycarbonate cannula (12 mm ID) containing multiple side holes was passed through an incision in the tip of the left ventricle and across the mitral valve into the left atrium. This venous cannula was secured with a ligature around the apex of the heart. A PE-200 catheter with multiple side holes was passed through the left atrial appendage to measure left atrial pressure, and another PE-200 catheter was threaded through the perfusion circuit to the end of the arterial cannula to measure pulmonary arterial pressure.

(1)

View larger version (35K): [in this window] [in a new window]   Fig. 1.   Schematic of the experimental setup. Pa, airway pressure; Ppa, pulmonary arterial pressure; Ppv, pulmonary venous pressure; CCD, charge-coupled device; A-D, analog-to-digital.

Videomicroscopy. A second thoracotomy was performed in the ninth left intercostal space, and a transparent window was implanted (21). After both thoracotomies were closed, the animal was placed in the right lateral decubitus position on a microscope stand, and the pneumothorax was evacuated. Evacuation of the pneumothorax stabilized the lung for videomicroscopy. The subpleural microcirculation under the window was observed with a modified Olympus BH2 reflectance microscope coupled to a Leitz Ultropak illuminator with a ×11 objective. Bright-field illumination through the illuminator was provided by a 200-W mercury arc lamp mounted on an optical bench. This light source was heavily filtered to prevent tissue damage with a combination of dichroic infrared-reflecting filters, broad band-pass ultraviolet-absorbing filters, and a narrow band-pass interference filter to illuminate the field only with the mercury green line (546 nm), a wavelength absorbed by hemoglobin (22). Illumination for fluorescence microscopy was provided by a 100-W mercury arc mounted on the sidearm of the BH2 microscope that was also filtered by dichroic infrared-reflecting filters and ultraviolet-absorbing filters. The light from this arc passed through a blue band-pass exciter filter (410-480 nm) and a high-pass dichroic mirror (cutoff wavelength 480 nm) that reflected the exciting light down through the objective onto the subpleural microcirculation beneath the window. Emitted light passed back through the objective, the dichroic mirror, and a yellow high-pass barrier filter (cutoff wavelength 510 nm). Video recordings of the subpleural microcirculation were made with a Sony SVHS video recorder (model SVO 5800) and an intensified charge-coupled device camera (Cohu model 5510 for fluorescent microscopy or Videoscope model 200E for bright-field microscopy) that was attached to the microscope with a zoom adapter.

Capillary recruitment recordings. In each lung, a microscopic field was selected that contained two to five clearly visible alveoli. The peristaltic pump was adjusted to produce a flow rate that recruited about one-third of the capillaries in the observed alveoli. During this period, the lung was conditioned by clamping the venous line briefly several times, which transiently elevated microvascular pressure enough to open essentially all the capillaries. After pressure and recruitment had returned to baseline, a 1-min video recording was made of capillary perfusion during steady flow and again during pulsatile perfusion (stroke volume 15-20 ml, rate 43-70/min). All recordings were made with the ventilator paused at end expiration. Because it has been demonstrated that capillary recruitment varies directly with arterial pressure but not with cardiac output (2), the pump flow rate and the height of the venous reservoir were adjusted so that the mean arterial and left atrial pressures during steady perfusion were the same as those observed during pulsatile perfusion. In four of the animals, steady-perfusion recordings preceded the pulsatile-flow recordings, and in the other three the order was reversed.

Microvascular indicator-dilution curves. In each lung, a microscopic field was selected that contained a single arteriole and a single venule of equal or smaller diameter. During observation of this field with fluorescence microscopy, test injections of dye [fluorescein isothiocyanate conjugated to 70-kDa dextran (Sigma Chemical), 20 mg/ml 0.9% saline] were made by using a loop just proximal to the lobar artery. Each limb of the loop contained a volume of ~10 ml and was controlled by a solenoid pinch valve (model NO/C-1367-92/93, Cole Parmer) on its downstream end. On one limb, the valve was open when deenergized, while on the other limb, the valve was closed. The normally closed limb was loaded with a bolus of dye. When the solenoids were energized, blood flow was diverted through the dye-containing limb, washing the dye into the left pulmonary artery. In this way, the bolus of dye was rapidly introduced into the arterial circulation without the pressure increase or movement of the microscopic field that occurs when high-pressure injections are made directly into the pulmonary artery. The passage of dye through the subpleural microcirculation was videotaped, and elapsed time in milliseconds was recorded on the videotapes by a time-date generator that was activated by the same switch that energized the solenoids of the injection loop.

Fluorescent microsphere injections. After the dye injections, fluorescent polystyrene microspheres (15 µm diameter; Molecular Probes) were injected under each flow condition into four of the seven animals (4, 5). The microspheres were ultrasonicated for 10 min and then vigorously vortexed just before injection. Blue-green microspheres (5 × 10⁵) were injected during steady flow, and red microspheres (5 × 10⁵) were injected during pulsatile flow. Microspheres were injected into the windkessel, which contained a stirring bar to ensure uniform mixing with the perfusate. A period of 10 min after each injection was allowed for all the microspheres to be washed from the windkessel. The injections did not cause any change in hemodynamic parameters. The order of pulsatile and steady flow during microsphere injection was the same as during the recruitment and transit time measurements.

Capillary recruitment measurements. We have used a binary definition of recruitment in which a segment was considered to be recruited if one or more red blood cells passed through that segment during a 1-min observation period (2, 6, 14, 15, 23, 24). This has been necessary, because in the 1-min recordings we made, we could tell only which segments were perfused at any moment in time, not the absolute level of blood flow in those segments. A problem with this measurement arises from the fact that the path of blood flow through the network of capillary segments frequently changes during a 1-min period (24). Therefore, the number of segments perfused at least once during a 1-min period is greater than the number perfused at any given moment; the greater the frequency of switching, the greater the difference. To obtain a more accurate measurement of recruitment at each moment in time, each 1-min recording was divided into six consecutive 10-s periods, and the level of recruitment during each 10-s period was measured. Observation periods <10 s were increasingly affected by the noise due to the natural temporal variation in the number of perfused segments (24) that obscured the change in recruitment resulting from the mode of perfusion. We reduced the noise in the recruitment measurements by averaging the six 10-s recruitment measurements from each single 1-min recording.

(2)

Capillary transit time measurements. Indicator-dilution curves were obtained by replaying the recordings of the injections and sampling image brightness at 30 Hz from rectangular areas over the arteriolar and venular lumens (vessel windows) and from areas over the adjacent alveoli (subtractor windows) by using a frame grabber board (model DT-2851, Data Translation) in a microcomputer (Dell 50-MHz 486). The sampling windows were movable and of adjustable size. The subtractor window signals were used to correct the vessel window signals for light emitted from the surrounding capillaries, as described in detail elsewhere (15). An average arteriolar and venular curve was obtained for each type of flow (steady or pulsatile) by averaging curves from six consecutive injections made during that flow condition. The baseline segment of each of these average curves before indicator entered the vessel was set to zero, and the tail of each curve (~5% of the area under the curve) was extrapolated to baseline as a monoexponential function. Finally, the area under each curve was set to unity. These curves served as the input and output functions for the determination of the distribution of capillary transit times by deconvolution by using damped least squares, as described previously (3). Mean capillary transit time () was calculated from the transport functions (the area of which was unity) as

(3)

where t₁ is the time when the intensity of the transport function curve (I) became greater than zero and t_n is the time when the curve returned to baseline (14, 15). The variance of the transit time (²) was calculated as

(4)

The coefficient of variation was computed as /. The minimum transit time (t_min) for each transport function was calculated as the time when 1% of the indicator had crossed the capillary network, and the maximum transit time (t_max) was the time when 99% of the indicator had crossed.

Statistics. We used the paired t-test to compare steady-flow values with the corresponding pulsatile-flow values. We accepted P < 0.05 as significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Capillary recruitment increased during pulsatile perfusion. The CPI doubled from 100 ± 11 to 205 ± 14 µm (P < 0.01; Table 1, Fig. 2). The number of perfused segments also doubled from 17 ± 2 to 33 ± 3 (P < 0.01; Table 1, Fig. 2). In the 38 alveoli we studied, we observed a sample of 33 capillary segments that were recruited only during pulsatile perfusion (Table 2). The perfusion of these segments was observed over a total of 1,956 pulsatile pump cycles. In 1,828 cycles (93.5%), capillary flow was present during systole and diastole. In the remaining 128 cycles (6.5%), flow was present during systole but not diastole (P < 0.01; Table 2).

View larger version (17K): [in this window] [in a new window]   Fig. 2.   Recruitment during steady flow compared with recruitment during pulsatile flow in 4 alveolar walls. Capillaries were crossed by 1 red blood cell in a 10-s period. Capillary perfusion index (CPI) was 104 µm during 10 s of steady flow and 192 µm during 10 s of pulsatile flow.

Mean transit time decreased from 5.2 ± 0.4 s during steady flow to 4.2 ± 0.5 s during pulsatile flow (P < 0.01; Table 3), and the transit time distribution narrowed (Fig. 3). The variance of the transit time distributions decreased from 9.1 ± 1.8 to 5.6 ± 1.3 s² (P < 0.01; Table 3), and maximum transit time decreased from 15.5 ± 1.5 to 12.3 ± 1.4 s (P < 0.01; Table 3), whereas the minimum transit time did not change significantly (P = 0.09; Table 3). The heterogeneity of the transit time distribution as expressed by the coefficient of variation decreased slightly from 0.56 ± 0.03 during steady flow to 0.53 ± 0.03 (P = 0.02; Table 3).

View larger version (18K): [in this window] [in a new window]   Fig. 3.   Transit time distribution during steady flow compared with that during pulsatile flow. Each curve is an average (n = 7 injections under each condition). During pulsatile flow, distribution narrowed (mean transit time and maximum transit time decreased without a significant change in minimum transit time) and became more homogeneous (variance and coefficient of variation decreased).

Microsphere injections demonstrated a slight upward redistribution of flow during pulsatile perfusion. While total lung blood flow was 1.3-fold higher during pulsatile flow (Table 4), there was a 1.7 ± 0.1-fold increase in the average blood flow in the plane where the microscopic observations were made (Fig. 4). Although there was a slight upward redistribution of flow, the correlation between steady flow and pulsatile flow to each piece was high (Fig. 4; average r² = 0.93 ± 0.04). In addition, spatial perfusion heterogeneity as measured by the coefficient of variation was 0.54 ± 0.04 with no difference between pulsatile and steady flow (P = 0.42; Table 3).

View larger version (29K): [in this window] [in a new window]   Fig. 4.   Correlation of pulsatile blood flow to steady blood flow within 613 pieces in a typical animal. Fluorescent microspheres were injected, and lung was sectioned into 1.9-cm3 pieces that could be grouped into isogravitational planes. Slope of the regression line through pieces in the uppermost 2 planes (red symbols, n = 13 pieces) was 1.83 (95% confidence limits = 1.45-2.21) compared with 0.91 (95% confidence limits = 0.76-1.06) for pieces in the most-dependent plane (yellow symbols, n = 15 pieces), indicating upward redistribution of flow. For the lung as a whole, correlation of pulsatile flow to steady flow for each 1.9-cm3 lung piece was high (r2 = 0.89).

At a constant mean arterial pressure of 7.5 ± 0.4 mmHg and left atrial pressure of 0.4 ± 0.2 mmHg, flow was 779 ± 68 ml/min during steady perfusion and 1,012 ± 67 ml/min during pulsatile perfusion (P < 0.01; Table 4). Thus pulmonary vascular resistance decreased from 0.63 ± 0.06 mmHg · ml¹ · s during steady perfusion to 0.48 ± 0.03 mmHg · ml¹ · s during pulsatile perfusion, a 24% decrease (P < 0.01; Table 4). Blood gas measurements at the beginning of the experiment were not significantly different from those at the end of the experiment (arterial PO₂ = 130 ± 5 Torr, arterial PCO₂ = 38 ± 2 Torr, pH = 7.38 ± 0.01, paired t-test, P > 0.20).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study shows that, at constant mean pulmonary arterial, left atrial, and airway pressures, pulsatile flow increased the number of recruited capillaries compared with steady flow. Although the systolic pressure rise during pulsatile flow was brief, capillaries recruited during systole did not derecruit during diastole. Recruitment was maintained in diastole, even though diastolic pressure was below the mean steady-flow pressure.

Hysteresis in the recruitment response was demonstrated by Johnson et al. (10, 18), who measured the diffusing capacity for carbon monoxide and showed that pulmonary capillary volume increased with exercise. After cessation of exercise, the fall in pulmonary capillary volume lagged the return of pulmonary arterial pressure to normal, showing that once pulmonary capillaries are opened, they tend to remain open for some time after pressure falls. Recently, we extended the findings of Johnson and colleagues by directly observing the capillaries. We found that capillaries recruit rapidly (<4 s) and derecruit somewhat less rapidly (~8 s) in response to rapid changes in perfusion pressure (8). In the present study, we have further defined the response time of the capillary bed, showing that capillaries recruit in <1 s, the duration of a single systole, but do not derecruit in such a brief time period (Table 2).

Capillary recruitment increases capillary volume. Because the capillary perfusion index doubled during pulsatile flow, the total length of perfused capillaries also doubled during pulsatile flow. If we assume that a doubling of perfused capillary length corresponds to an equal change in capillary volume, then capillary volume doubled during pulsatile flow. This estimated change in capillary volume assumes that capillaries newly recruited during pulsatile perfusion were also fully distended. We can test this assumption by calculating the capillary volume change from our independent measurements of capillary transit time and blood flow by using the following equations

(5)

From this relationship, the relative change in capillary volume can be calculated as

(6)

In the present study, capillary transit time during pulsatile flow was 81% of the steady-flow capillary transit time. The average blood flow in the plane of the microscopic observations increased 1.7-fold during pulsatile flow as measured by microspheres. From these changes, we calculate a 1.4-fold increase in capillary volume during pulsatile flow

(7)

Because the 1.4-fold increase calculated from the transit time and blood flow measurements is less than the doubling predicted from the change in perfused capillary length, we conclude that the capillaries newly recruited during systole were not fully distended. Although these capillaries were not fully distended, they were perfused throughout the pulsatile cycle 93.5% of the time (Table 2).

Our results assume that measurements of recruitment and transit time in the subpleural microcirculation parallel those in the interior of the lung. Recent work by Short et al. (17) has shown interior and subpleural recruitment patterns to be similar. They found that the decreased density of subpleural capillaries in the rat lung resulted in recruitment measurements that were about one-half the magnitude of those for interior capillaries. However, changes in recruitment of subpleural capillaries accurately reflected changes in recruitment of interior capillaries over a range of pressures that spanned low zone 3 to high zone 1. Therefore, the absolute values of the recruitment measurements we report are likely to be less than those for interior capillaries, but the relative change in recruitment with the mode of perfusion should be the same. Although it is less clear how morphological differences between subpleural and interior capillaries affect transit time, a study by Capen et al. (1) in dogs showed that subpleural plasma capillary transit times measured by dye dilution were similar to red blood cell capillary transit times for the whole lung measured by carbon monoxide-diffusing capacity.

Rather than maintaining constant flow when changing between pulsatile and steady perfusion, we increased flow during pulsatile perfusion to maintain the mean arterial and venous pressures measured during steady flow (Table 4). We did this because it has been demonstrated that capillary recruitment varies directly with pressure but not with flow. Observing the microcirculation in the upper lung, Capen and Wagner (2) measured an increase in recruited capillaries when pulmonary arterial pressure was increased by airway hypoxia. They also measured a simultaneous increase in cardiac output. However, when pulmonary arterial pressure was decreased by infusion of prostaglandin E₁, capillaries derecruited while cardiac output increased further. They concluded that capillary recruitment was dependent on increased pulmonary arterial pressure, rather than increased cardiac output alone.

In addition to increased capillary recruitment and decreased transit time, we observed an upward redistribution of flow (Fig. 4) during pulsatile perfusion that we believe was due to distribution of the increased flow to the least-recruited areas. This result is in agreement with that of Capen and Wagner (2), who showed an upward redistribution of blood flow and capillary recruitment during hypoxic pulmonary hypertension. In a directly relevant study, Maloney et al. (12) also observed an upward redistribution of blood flow in isolated dog lungs when flow was changed from steady to pulsatile. They hypothesized that the systolic pressure rise exceeded the opening pressure of collapsible vessels in the upper lung that were otherwise closed. The techniques used by these investigators to measure blood flow did not have the necessary spatial resolution to determine the size of the vessels that were opened by the change to pulsatile flow and the time resolution that was necessary to determine whether the vessels remained perfused during diastole.

Consistent with work by Glenny et al. (5), the redistribution of flow that we observed was a small fraction of the total variation of flow within each gravitational plane. The spatial heterogeneity of flow (coefficient of variation = 0.54 ± 0.04; Table 3) was also similar to previously reported values for the dog lung (5). Interestingly, despite increased flow rates during pulsatile flow and the presence of a pulsatile waveform, the heterogeneity of perfusion was constant and the correlation between pulsatile and steady flow to each 1.9-cm³ piece was high (Fig. 4). This suggests a relatively fixed vascular structure that determines the distribution of perfusion independent of flow rate and pulsatile waveform.

Our results are in agreement with the study of Johnson et al. (9), who found a lower pulmonary vascular resistance during pulsatile flow in blood-perfused newborn lamb lungs. Similarly, Raj et al. (16), who studied isolated rabbit lungs perfused alternately with steady and pulsatile flow, measured a 30% drop in pulmonary vascular resistance during pulsatile flow. This decrease is close to the 24% decrease we measured in this study. Using micropuncture to measure the pressure in 20- to 50-µm arterioles and venules, they found that the fall in resistance that occurred during pulsatile flow was due to a fall in microvascular resistance. Our measurements show a mechanism that can explain the fall in resistance: the doubling of capillary recruitment would lead to a significant decrease in microvascular resistance.

In an elegant study, Lee and DuBois (11) showed that capillary blood flow was pulsatile. They measured the uptake of nitrous oxide into pulmonary capillary blood inside a body plethysmograph. This gas, which because of its high solubility has a very rapid uptake across the alveolar-capillary membrane, caused a pulsatile fluctuation in the plethysmographic signal. This experiment could not determine, however, whether pulmonary capillary flow pulsations were simply pulsatile red cell velocity changes or whether pulsatile changes in capillary volume and surface area also occurred. In a similar experiment, Menkes et al. (13) demonstrated pulsatile pulmonary capillary uptake of carbon monoxide. This result indicated that pulsatile changes in capillary volume occurred but could not rule out whether pulsatile changes in red cell velocity were also occurring. The distinction is important, because a pulsatile change in red cell velocity in the absence of a change in capillary volume increases the heterogeneity of capillary red cell transit times with the potential for those cells with the fastest transit times to cross the capillaries before they are completely saturated with oxygen (19, 20). This potentially detrimental change would be offset by a pulsatile change in capillary volume, causing red cell transit times to become more homogeneous. Although we observed pulsatile changes in red cell velocity in some capillaries,¹ we also found that capillaries were recruited (Tables 1 and 2). The overall result was that the distribution of red cell transit times became more homogeneous. Mean capillary transit time, maximum capillary transit time, the variance of the distribution, and the coefficient of variation of the distribution decreased, whereas the minimum transit time did not change (Table 3, Fig. 3). In addition to a more homogeneous transit time distribution, capillary recruitment adds directly to gas exchange surface area. Although we did not measure the response time of blood gas changes after a change in inspired gas concentrations, Hauge and Nicolaysen (7) showed that the rate of rise of PO₂ in pulmonary venous blood after an increase in the inspired oxygen concentration was greater during pulsatile flow than during steady flow. On the basis of these studies and our direct measurements of transit time and capillary recruitment, we conclude that gas exchange is more efficient during pulsatile flow.

In summary, we observed an increase in capillary recruitment during pulsatile flow. Most of the capillaries recruited by the systolic pulse were perfused throughout the pulsatile cycle. Our data support the idea that recruitment of capillaries by pulsatile flow increases the cross-sectional area of the vascular bed, lowering resistance and increasing gas exchange surface area.