Journal of Applied Physiology
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

J Appl Physiol 82: 1668-1676, 1997;
8750-7587/97 $5.00
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF) Free
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Topulos, G. P.
Right arrow Articles by Butler, J. P.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Topulos, G. P.
Right arrow Articles by Butler, J. P.

Journal of Applied Physiology
Vol. 82, No. 5, pp. 1668-1676, May 1997
PULMONARY CIRCULATION AND LUNG FLUID BALANCE

Fractional changes in lung capillary blood volume and oxygen saturation during the cardiac cycle in rabbits

George P. Topulos1,2, Nina R. Lipsky1, John L. Lehr1, Rick A. Rogers1, and James P. Butler1

1 Physiology Program, Harvard School of Public Health, and 2 Department of Anesthesia, Brigham and Women's Hospital, Boston, Massachusetts 02115

ABSTRACT
INTRODUCTION
METHODS
RESULTS AND DISCUSSION
ACKNOWLEDGEMENTS
FOOTNOTES
REFERENCES

ABSTRACT

Topulos, George P., Nina R. Lipsky, John L. Lehr, Rick A. Rogers, and James P. Butler. Fractional changes in lung capillary blood volume and oxygen saturation during the cardiac cycle in rabbits. J. Appl. Physiol. 82(5): 1668-1676, 1997.---Changes in local pulmonary capillary blood volume (Vc) and oxygen saturation (S) have been difficult to measure in live animals. By utilizing the differences in absorption of light at two wavelengths (650 and 800 nm), we estimated the fractional change in Vc and S during the course of the cardiac cycle in eight anesthetized, ventilated rabbits at low and high lung volumes. Observations were made of the pattern of diffusely backscattered light, from an ~1-cm3 volume of lung illuminated with a point source placed on the pleural surface through a thoracotomy. At low lung volume, the fractional change in Vc was ~13%, the change in S was ~4.6%, and the mean S was close to 77%. The fluctuations in Vc and S lagged behind peak systemic blood pressure by about one-fifth and three-fifths of a cycle, respectively. At high lung volume, there were no important fluctuations in Vc or S, and the mean S was ~82%. These results are consistent with fluctuations in pulmonary capillary pressure and gas exchange over the cardiac cycle, and with decreasing capillary compliance with increasing lung volume.

gas exchange; microcirculation; diffuse light scattering

INTRODUCTION

A NUMBER OF FEATURES of the lung that are crucial to its function are also difficult or heretofore impossible to measure in vivo. These include pulmonary surface area-to-volume ratio (A/V) [i.e., the area of air-tissue interface per unit absolute volume (tissue plus gas)], capillary blood volume (Vc), and microcirculatory oxygen saturation (S). A/V and Vc are critical determinants of lung function. Capillary compliance is an important determinant of the pattern of pulmonary perfusion, which in turn influences red cell transit times and gas exchange, as well as leukocyte transit times and sequestration, which are important in the inflammatory response. The interaction of the deformability of the leukocytes and the capillary bed through their respective compliances thus requires measurements of both. Cellular deformability is an active area of current research, but there are few data on the compliance of the capillary bed in vivo. Finally, S is an important measure of how well the lung is functioning as a gas exchanger.

We sought to overcome these measurement limitations by exploiting the characteristics of light diffusely scattered by the lungs. These characteristics depend on both geometric features such as A/V and Vc, as well as on functional parameters such as S. In this paper, we report on a dual-wavelength extension of previous diffuse light-scattering (DLS) techniques (3, 10, 17), that permit measurements of changes in these specific physiological parameters. The variations in Vc and S during the cardiac cycle in normal animals at rest are expected to be small and potentially of only moderate physiological significance. On the other hand, they provide a useful test of the resolving power of this new technique. In addition, there are no direct measures of these variables in the literature, and if they had, in fact, been quantitatively significant, it would have been both surprising and important. These same considerations apply to the comparisons of mean saturation (<OVL>S</OVL>) at high and low lung volumes. We believe it is critical to establish and report that such measurements can, in fact, be made to establish their feasibility in the lower bound range of changes in physiologically important variables. Having done so, it opens the window for future investigations where large fluctuations in, e.g., Vc may be anticipated, such as with pulmonary hypertension or where changes in lung recoil are associated with pathophysiological conditions. Similarly, examination of changes in S by this technique may reveal differences in oxygen transport in normal vs. abnormal lungs through the kinetics of oxygen transport.

METHODS

Glossary

Physiological variables
A/V Pulmonary surface area-to-volume ratio, cm-1
Vc Capillary blood volume, ml
S Microcirculatory oxygen saturation
 <OVL>S</OVL> Mean microcirculatory oxygen saturation over the cardiac cycle
SPO2 Arterial oxygen saturation from pulse oximetry
T Mean alveolar septal thickness, cm
[Hb] Total hemoglobin concentration, mmol/l

Optics and morphometry
Pi Probability of absorption by species i
Lm Geometric or morphometric mean linear intercept, cm
Lopt Optical mean free path, cm
 alpha 3Lm/2Lopt
 epsilon i,lambda Extinction coefficient for species i at wavelength lambda , cm-1 · l· mmol-1
[Ci] Concentration of absorbing species i, mmol/l
Anb Light absorption by nonblood lung tissue, cm-1

Optical measurements and derived quantities
I Photon flux density (intensity), photons · s-1 · cm-2
r Radial distance on the pleural surface from the optical center, cm
kdiff,lambda Diffuse extinction coefficient at wavelength lambda , cm-1
l Photometer cuvette width (= 1 cm), cm
OD  epsilon [C]l, Optical density
f [l/(1 - hematocrit)](partial k2diff,800 /partial OD), cm-1

Parameters of sinusoidal variation
x Fraction of time in the cardiac cycle, measured from peak systemic blood pressure
M Magnitude or amplitude (1/2 peak to peak) of the sinusoidal variation in kdiff over the cardiac cycle, cm-1
 phi Phase shift relative to peak systemic blood pressure

Point illumination of the pleural surface of the lung results in a distinctive pattern of backscattered light. Deriving the values of physiological parameters from this pattern involves several steps: 1) animal preparation, 2) delivering the light to the lung, 3) capturing and quantifying the pattern of backscattered light, 4) deriving the diffuse absorption coefficient (kdiff) from the pattern of backscattered light, and 5) calculating the values of the physiological parameters of interest from kdiff and calibration data. The following paragraphs describe each of these procedures.

Animal preparation. New Zealand White rabbits weighing 3-3.8 kg were anesthetized with ketamine (50 mg/kg) and xylazine (1-2 mg/kg) via intramuscular injection. A percutaneous intravenous catheter was placed in an ear vein. Anesthesia was maintained with pentobarbital sodium (6-12 mg iv) as needed to prevent spontaneous movement and response to paw pinch. A tracheotomy was performed below the larynx, and the trachea was cannulated with a short tube. The lungs were ventilated with a respirator (model 645, Harvard Apparatus, South Natick, MA) at a rate of 40 breaths/min, with a tidal volume of 10 ml/kg and with an end-expiratory pressure of 3-5 cmH2O. An arterial catheter was placed in the carotid artery via the tracheotomy incision. The animal was placed in the right lateral position, a left thoracotomy was performed, the parietal pleura was opened, and the ribs were spread, exposing an ~1.5 × 3-cm area of the middle or lower lobe. The interspace was chosen such that no lobar margins were visible through the thoracotomy.

Arterial saturation was estimated (SpO2) with a pulse oximeter (model 3700, Ohmeda) placed on the tongue. Pressure at the airway opening (Pao) was measured with a pressure transducer (Validyne MP-45, ±50 cmH2O diaphragm), calibrated over its full range of use with a water manometer. Blood pressure (BP) was measured with a pressure transducer (Transpac II, Abbott) and monitor (model 414, Tektronix), calibrated over its full range of use with a mercury manometer. Pao, BP, SpO2, and time of the laser trigger (see Delivering light to the lung) were recorded by using a computer data-acquisition system and 14-bit analog-to-digital board (Codas, Dataq Instruments, Akron, OH) sampling each channel at 600 or 750 Hz.

Delivering light to the lung. An optical fiber delivered light to the lung surface. A rubber cylinder (the plunger tip of a tuberculin syringe, ~5-mm diameter and 7-mm tall) with an optical fiber (0.75-mm diameter, ESKA acrylic, Edmund Scientific) passed through its center was cemented with cyanoacrylate to the visceral pleura of the lung via the thoracotomy. The rubber cylinder held the fiber in place approximately perpendicular to the lung surface and prevented camera overload. The fiber was supported above the animal on a flexible gooseneck stand, so it exerted minimal force on the lung surface. The light source was a tunable pulsed dye laser (VSL-337ND nitrogen laser and DLM-337220 dye laser module, Laser Science, Cambridge, MA), tuned to either 650 nm [spectrophotometric-grade DCM dye in dimethyl sulfoxide (DMSO)] or 800 nm (spectrophotometric-grade DOTC dye in DMSO). Each laser pulse was 3 ns in duration and, depending on wavelength, 10-30 µJ in total energy (higher for 650 nm and lower for 800 nm secondary to efficiency of the dyes); on a given day the energy at a given wavelength varied only slightly within that range. At any given time the laser was configured to deliver either 800- or 650-nm light. Because of the time involved in changing the laser output configuration, data for each animal were first collected at 800 nm, then the dye and laser grating were changed to yield 650-nm light, and then the remaining data were obtained.

Capturing the pattern of backscattered light. The pattern of backscattered light observed on the pleural surface was captured by a charge-coupled device (CCD) camera (CE200A camera electronics unit and CH250 thermoelectrically cooled camera head, Photometrics, and TH7896A 1024 × 1024 pixels CCD, Thompson), lens (Nikon AF 50 mm 1:1.8), and video software (Image 200, Photometrics). Photographs were ~0.75 × 1.5 cm, with ~50 pixels/cm. This discretization scale varied with the exact distance of the lung from the camera. The camera shutter (open for 30 ms) was controlled so each image captured the light from a single laser pulse. The images were digitized by a 14-bit analog-to-digital converter and stored on a computer (Gateway 386) for later analysis. We note that this technique sums the photons at any pixel location in the CCD camera regardless of the time of arrival within the 30-ms interval when the shutter was open. Because of the cumulative nature of this method, any variations in arrival time due to tissue optical capacitance, pulse dispersion, and especially time-of-flight distributions due to the random walk character of photon migration can be ignored.

Quantifying the pattern of backscattered light. The images from the CCD camera were analyzed to yield values of the intensity of backscattered light (I) as a function of radial distance (r) from the optical center, as described elsewhere (10). I values were corrected for background light (i.e., the signal collected when the laser input to the optical fiber was blocked) and for dark current and nonuniformities caused by variations in the photosensitive surface (i.e., the signal detected with the CCD camera lens capped). Each picture set consisted of 15 images and 1 or 2 background pictures taken with the laser input to optical fiber blocked (described in Experimental protocol).

Estimating kdiff from the pattern of backscattered light. In the past, we have shown that for light diffusely scattered from a point source the intensity I at distance r from the origin is proportional to (Lopt/r3)e-kdiffr, where Lopt is the optical mean free path (roughly the average distance a photon travels before it is scattered or absorbed). It follows that ln[I(r)r3] is proportional to -kdiffr + ln(Lopt) and that, therefore, the slope of the quantity ln[I(r)r3] plotted against r is a direct measure of -kdiff (10).

Calculating the values of the physiological parameters of interest from kdiff. Calculating the values of the physiological parameters of interest from kdiff is based on optical theory and calibration. The Lambert-Beer law states that when uniformly illuminating light passes through a purely absorptive medium (i.e., without scattering)1, its intensity falls off exponentially with distance (Lambert) and that the extinction coefficient depends linearly on the concentration of absorbing species (Beer). In a medium in which light is not only absorbed but also diffusely scattered, the light intensity also falls off exponentially with distance. However, kdiff, rather than depending linearly on the concentration of absorbing species as in the Beer law, satisfies the following equation (11)
<IT>k</IT><SUP>2</SUP><SUB>diff</SUB> = (3/<IT>L</IT><SUP>2</SUP><SUB>opt</SUB>) <LIM><OP>∑</OP><LL><IT>i</IT></LL></LIM> <IT>P</IT><SUB><IT>i</IT></SUB> (1)
where the summation is over all absorbing species (i), and Pi is the probability of absorption, i.e., the fraction absorbed divided by the fraction absorbed or scattered.

For light transmission through very thin layers, such as alveolar septa, the probability of absorption is equal to the product of the mean septal thickness (T), the molar extinction coefficient (epsilon ) and the concentration [C] of the absorbing species ([Ci]); this follows from the approximation to exponential decrease for small distances. In this approximation, Pi = epsilon i[Ci]T. Next, assume that T and mean transverse capillary thickness are equal and that mean septal area (A) and mean capillary area are also equal (these assumptions are not critical; we will see below (Eqs. 12, 15, 19) that any error introduced here will cancel out in the final formula). It follows that Vc is the product of T and corresponding A. Thus blood volume density is given by TA/V, where V is total volume, and we may substitute for T to obtain the absorption probabilities in the form
<IT>P</IT><SUB><IT>i</IT></SUB> = &egr;<SUB><IT>i</IT></SUB>[C<SUB><IT>i</IT></SUB>] <FR><NU>Vc</NU><DE>V</DE></FR> <FR><NU>V</NU><DE><IT>A</IT></DE></FR> (2)
Similarly, because we have assumed that capillary bed area is equal to septal surface area, we may use the fact that V/A (the inverse of the pulmonary surface area-to-volume ratio) is known from classic stereology (21) to be related to the geometric or morphometric mean linear intercept (Lm) by V/A = Lm/2. This yields
<IT>P</IT><SUB><IT>i</IT></SUB> = &egr;<SUB><IT>i</IT></SUB>[C<SUB><IT>i</IT></SUB>] <FR><NU>Vc</NU><DE>V</DE></FR> <FR><NU><IT>L</IT><SUB>m</SUB></NU><DE>2</DE></FR> (3)
Departures from equality of capillary bed area and septal surface area will change the factor of 2 in this relationship, but, again, this will not affect our final results. Let k2i,diff = (3/L2opt)Pi. This is the contribution from species i to k2diff. From Eq. 3 we get
<IT>k</IT><SUP>2</SUP><SUB>i,diff</SUB> =  &agr; <FR><NU>1</NU><DE><IT>L</IT><SUB>opt</SUB></DE></FR> &egr;<SUB><IT>i</IT></SUB>[C<SUB><IT>i</IT></SUB>] <FR><NU>Vc</NU><DE>V</DE></FR> (4)
where alpha  = 3Lm/2Lopt. alpha  is a measure of the departure of Lm from Lopt. Lopt is greater than Lm because septa are largely transparent; that is, most photons that hit a septum go through without being absorbed or scattered. Furthermore, alpha  varies with lung volume because the role that scattering plays is lung volume dependent, and alpha  is known to increase as lung volume decreases. Specifically, alpha  ranges from 0.56 to 0.26 as lung volume ranges from 40 to 100% total lung capacity, according to the data of Suzuki et al. (17), but is constant at any given lung volume. We note that only in the simplest models of scattering is alpha  approximately constant. Especially problematic is the lung volume-dependent anisotropy of elastic photon scattering from septal surfaces, which is the dominant optical event in the lung. On the other hand, as shown below (Eqs. 12, 15, 19), we do not need to estimate or measure alpha itself, because it will ultimately cancel out of all calculations relating the primary measurements to the physiological quantities of interest.

In any particular circumstance, Eq. 4 must be used for each absorbing species. In our experiments there are three naturally occurring absorbers: nonblood components of lung tissue, oxygenated hemoglobin, and nonoxygenated hemoglobin. For purposes of calibration, there is one artificial absorber indocyanine green dye (ICG). Expanding Eq. 1, by using Eq. 4, to include a term for each of these species, we have the relationship between kdiff (which is derived directly from our optical measurements) and several physiological factors of interest
<IT>k</IT><SUP>2</SUP><SUB>diff</SUB> = <FR><NU>&agr;</NU><DE><IT>L</IT><SUB>opt</SUB></DE></FR> <FENCE>A<SUB>nb</SUB> + [Hb] <FR><NU>Vc</NU><DE>V</DE></FR> [&egr;<SUB>HbO<SUB>2</SUB></SUB>S + &egr;<SUB>Hb</SUB>(1 − S)]</FENCE>
<FENCE>+ &egr;<SUB>ICG</SUB>(1 − Hct) <FR><NU>Vc</NU><DE>V</DE></FR> [ICG]</FENCE> (5)

where Anb is the contribution to absorption from nonblood components (e.g., lung tissue), [Hb] is the total hemoglobin concentration in millimoles per liter, epsilon HbO2 and epsilon Hb are the wavelength-dependent molar extinction coefficients in liters per millimole centimeter for oxygenated and nonoxygenated hemoglobin, respectively2, epsilon ICG is the wavelength-dependent molar extinction coefficient for ICG, and [ICG] its concentration. The fact that ICG is only distributed in plasma accounts for the additional factor (1 - Hct), where Hct is the hematocrit, in the last term, which represents the contribution to kdiff from ICG.

We will now investigate Eq. 5 under several different circumstances to successively solve for the fractional change in Vc (delta Vc/Vc), the change in S over the cardiac cycle (delta S), and mean microcirculatory oxygen saturation (<OVL>S</OVL>) over the cardiac cycle. It is important to recognize that we have lumped all of the blood in the pulmonary circulation in the 1- to 2-cm roughly hemispherical volume sampled by our technique under the phrase "capillary blood." This is only approximately true, but more than two-thirds of the blood in rabbit parenchyma is in pulmonary capillaries, vessels <20-µm diameter (4). Furthermore, the contribution of capillary blood to our measurements is likely to be much larger because we are sampling near the pleural surface, where large and even moderate-sized vessels are scarce.

Solving for delta Vc/Vc. Light of 800 nm is equally absorbed by nonoxygenated and oxygenated hemoglobin, that is, epsilon HbO2 800 = epsilon Hb 800, and at this isobestic wavelength3 Eq. 5 reduces to
<IT>k</IT><SUP>2</SUP><SUB>diff 800</SUB>
= <FR><NU>&agr;</NU><DE><IT>L</IT><SUB>opt</SUB></DE></FR> <FENCE>A<SUB>nb</SUB> + [Hb]&egr;<SUB>Hb 800</SUB> <FR><NU>Vc</NU><DE>V</DE></FR> + (1 − Hct)&egr;<SUB>ICG 800</SUB> <FR><NU>Vc</NU><DE>V</DE></FR> [ICG]</FENCE> (6)

If we fix Anb, Vc, and [Hb], we see that k2diff is linearly related to [ICG], and the slope of this relationship is given by
<FR><NU>∂<IT>k</IT><SUP>2</SUP><SUB>diff 800</SUB></NU><DE>∂[ICG]</DE></FR> = <FR><NU>&agr;</NU><DE><IT>L</IT><SUB>opt</SUB></DE></FR> (1 − Hct)&egr;<SUB>ICG 800</SUB> <FR><NU>Vc</NU><DE>V</DE></FR> (7)
Experimentally we approximated a constant Anb by holding lung volume constant, a constant [Hb] by making all measurements of k2diff over a short period of time, and a constant Vc by averaging all measurements of k2diff over the cardiac cycle.

By contrast, when [ICG] = 0, the variation in k2diff is related to the changes in Vc by
&dgr;<IT>k</IT><SUP>2</SUP><SUB>diff 800</SUB> = <FR><NU>&agr;</NU><DE><IT>L</IT><SUB>opt</SUB></DE></FR> [Hb]&egr;<SUB>Hb 800</SUB> <FR><NU>&dgr;Vc</NU><DE>V</DE></FR> (8)
Combining Eq. 8 with Eq. 7 to eliminate the unknown constant alpha /Lopt, and rearranging, we get
<FR><NU>&dgr;Vc</NU><DE>Vc</DE></FR> = <FR><NU>(1 − Hct)&egr;<SUB>ICG 800</SUB></NU><DE>[Hb]&egr;<SUB>Hb 800</SUB></DE></FR> <FENCE><FR><NU>∂<IT>k</IT><SUP>2</SUP><SUB>diff 800</SUB></NU><DE>∂[ICG]</DE></FR></FENCE><SUP>−1</SUP> &dgr;<IT>k</IT><SUP>2</SUP><SUB>diff 800</SUB> (9)

Thus experiments designed to assess the variation in Vc consist of two pieces. First, there is the estimate of the change in k2diff with change in [ICG], that is, partial k2diff 800/partial [ICG]. This constitutes a calibration insofar as increased absorption of 800-nm light secondary to an ICG infusion is optically equivalent to changes in absorption secondary to an increased blood volume or Hct. Strictly speaking, this equivalence must be understood in the context that ICG is present in the plasma and not in red blood cells; this, however, is accounted for by the presence of the factor (1 - Hct) in Eqs. 5 and 6. Furthermore, we can measure changes in k2diff 800 and independently measure the changes in light absorption, as a result of the ICG infusion, by spectrophotometry of blood samples. Second, we measure cyclic changes in k2diff 800 over the cardiac cycle in the absence of dye. Note that the use of Eqs. 7 and 8 allows us to eliminate unknown constants, at the expense of being able to estimate delta Vc /Vc rather than Vc itself.

Experimentally, optical measurements of k2diff 800 are compared with spectrophotometric measurements of serum optical density (OD) at 800 nm. The OD is given by
OD = OD<SUB>0</SUB> + &egr;<SUB>ICG 800</SUB>[ICG]<IT>l</IT> (10)
where OD0 is the optical density of the serum with no dye present and l is the cuvette width (1 cm) (20). The calibration term that appears in Eq. 9 can then be written
<FR><NU>1</NU><DE>(1 − Hct)&egr;<SUB>ICG 800</SUB></DE></FR> <FENCE><FR><NU>∂<IT>k</IT><SUP>2</SUP><SUB>diff 800</SUB></NU><DE>∂[ICG]</DE></FR></FENCE> = <FR><NU><IT>l</IT></NU><DE>1 − Hct</DE></FR> <FR><NU>∂<IT>k</IT><SUP>2</SUP><SUB>diff 800</SUB></NU><DE>∂OD</DE></FR> ≡ <IT>f</IT> (11)
The slope of optical measurements of k2diff 800 regressed against photometrically determined values of OD weighted by (1 - Hct) is then an estimate of the calibration factor f, needed to convert optical measurements into estimates of variations in physiological variables. With this calibration factor, Eq. 9 becomes simply
<FR><NU>&dgr;Vc</NU><DE>Vc</DE></FR> = <FR><NU>1</NU><DE><IT>f</IT>[Hb]&egr;<SUB>Hb 800</SUB></DE></FR> &dgr;<IT>k</IT><SUP>2</SUP><SUB>diff 800</SUB> (12)

Solving for <OVL><IT>S</IT></OVL>. <OVL>S</OVL> can be obtained similarly. Subtracting the average value of k2diff at 800 and 650 nm, using Eq. 5 with [ICG] = 0, yields
<OVL><IT>k</IT><SUP>2</SUP></OVL><SUB>diff 800</SUB> − <OVL><IT>k</IT><SUP>2</SUP></OVL><SUB>diff 650</SUB>
= <FR><NU>&agr;</NU><DE><IT>L</IT><SUB>opt</SUB></DE></FR> [Hb] <FR><NU>Vc</NU><DE>V</DE></FR> [&egr;<SUB>Hb 800</SUB> − &egr;<SUB>HbO<SUB>2</SUB> 650</SUB><OVL>S</OVL> − &egr;<SUB>Hb 650</SUB>(1 − <OVL>S</OVL>)] (13)

where the overbar denotes the average over the cardiac cycle. Dividing by the calibration relationship given in Eq. 7 and the definition in Eq. 11, we find
<OVL><IT>k</IT><SUP>2</SUP></OVL><SUB>diff 800</SUB> − <OVL><IT>k</IT><SUP>2</SUP></OVL><SUB>diff 650</SUB>
= <IT>f</IT>[Hb][&egr;<SUB>Hb 800</SUB> − &egr;<SUB>HbO<SUB>2</SUB> 650</SUB><OVL>S</OVL> − &egr;<SUB>Hb 650</SUB>(1 − <OVL>S</OVL>)] (14)

Solving for <OVL>S</OVL> gives the final formula
<OVL>S</OVL> = <FR><NU>1</NU><DE>&egr;<SUB>Hb 650</SUB> − &egr;<SUB>HbO<SUB>2</SUB> 650</SUB></DE></FR> <FENCE><FR><NU><OVL><IT>k</IT><SUP>2</SUP></OVL><SUB>diff 800</SUB> − <OVL><IT>k</IT><SUP>2</SUP></OVL><SUB>diff 650</SUB></NU><DE><IT>f</IT>[Hb]</DE></FR> − &egr;<SUB>Hb 800</SUB> + &egr;<SUB>Hb 650</SUB></FENCE> (15)

Solving for delta S. Finally, we turn to the question of what we can learn about changes in S during the cardiac cycle. In the absence of ICG and by using 650-nm wavelength light, Eq. 5 reduces to
<IT>k</IT><SUP>2</SUP><SUB>diff 650</SUB> = <FR><NU>&agr;</NU><DE><IT>L</IT><SUB>opt</SUB></DE></FR> <FENCE>A<SUB>nb</SUB> + [Hb] <FR><NU>Vc</NU><DE>V</DE></FR> [&egr;<SUB>HbO<SUB>2</SUB> 650</SUB>S + &egr;<SUB>Hb 650</SUB>(1 − S)]</FENCE> (16)

At fixed lung volume, variations in k2diff at 650 nm arise from variations in both Vc and S. Thus
&dgr;<IT>k</IT><SUP>2</SUP><SUB>diff 650</SUB> = <FR><NU>&agr;</NU><DE><IT>L</IT><SUB>opt</SUB></DE></FR> [Hb] <FENCE>[&egr;<SUB>HbO<SUB>2</SUB> 650</SUB>S + &egr;<SUB>Hb 650</SUB>(1 − S)] <FR><NU>&dgr;Vc</NU><DE>V</DE></FR></FENCE>
<FENCE>+ <FR><NU>Vc</NU><DE>V</DE></FR> (&egr;<SUB>HbO<SUB>2</SUB> 650</SUB> − &egr;<SUB>Hb 650</SUB>)&dgr;S</FENCE> (17)

Use of Eqs. 7 and 11, again to eliminate the unknown alpha /Lopt, yields
&dgr;<IT>k</IT><SUP>2</SUP><SUB>diff 650</SUB> = [Hb] <IT>f</IT>
· <FENCE>[&egr;<SUB>Hb 650</SUB>(1 − S) + &egr;<SUB>HbO<SUB>2</SUB> 650</SUB>S] <FR><NU>&dgr;Vc</NU><DE>Vc</DE></FR> + (&egr;<SUB>HbO<SUB>2</SUB> 650</SUB> − &egr;<SUB>Hb 650</SUB>)&dgr;S</FENCE> (18)

In this equation, the first term within the braces depends in part on delta Vc/Vc, which has already been estimated by Eq. 12. Substituting for delta Vc/Vc, and using the average value of S determined above in Eq. 15, we have
&dgr;S = <FR><NU>1</NU><DE><IT>f</IT>[Hb](&egr;<SUB>HbO<SUB>2</SUB> 650</SUB> − &egr;<SUB>Hb 650</SUB>)</DE></FR>
· <FENCE>&dgr;<IT>k</IT><SUP>2</SUP><SUB>diff 650</SUB> − <FR><NU>&egr;<SUB>Hb 650</SUB>(1 − <OVL>S</OVL>) + &egr;<SUB>HbO<SUB>2</SUB> 650</SUB><OVL>S</OVL></NU><DE>&egr;<SUB>Hb 800</SUB></DE></FR> &dgr;<IT>k</IT><SUP>2</SUP><SUB>diff 800</SUB></FENCE> (19)

The effects of changes in various physiological variables on the optical measurements may be summarized as is done in Table 1.

Table 1. Effects of changes in physiological variables on optical measurements

If  Then  And Therefore 
 up-arrow  A/V  Lm and Lopt down-arrow I(r) falls off faster with r and kdiff up-arrow  
 up-arrow Hct or up-arrow  Vc Absorption up-arrow  at 650 and 800 nm kdiff up-arrow  at both 650 and 800 nm
 down-arrow O2 saturation Absorption up-arrow  at 650 nm kdiff up-arrow  only at 650 nm
 down-arrow Lung volume A/V up-arrow Lm and Lopt down-arrow , kdiff up-arrow  at both 650 and 800 nm
A/V, pulmonary surface area-to-volume ratio; Lm geometric or morphometric mean linear intercept; Lopt, optimal mean free path; I, photon flux density (intensity); r, radial distance on pleural surface from optical center; kdiff, diffuse extinction coefficient; Hct, hematocrit; Vc, capillary blood volume; up-arrow , increase; down-arrow , decrease.

Experimental protocol. Each data set consisted of a sequence of 15 photographs and the simultaneously recorded Pao, BP, SpO2, and laser trigger marker, all vs. time. Data were collected by using one of two protocols, which differed only in the lung volume history at the time the data were collected. In protocol I (4 animals), a data set was collected during a short breath hold at a transpulmonary pressure of 5 or 15 cmH2O (representing nominally end expiration and end inspiration, respectively), after a period of normal ventilation. Between breath holds the rabbit was allowed to recover; SpO2 and BP returned to baseline, and additional anesthetic was given as needed. The sequence of 15 photographs was taken over a period of a few seconds, without synchrony between the picture frequency and the heart rate. This protocol was repeated four times at each transpulmonary pressure and at each wavelength, for a total of 64 possible data sets.

In protocol II (4 animals), a data set was collected during normal ventilation by triggering the laser from the ventilator cam position at either end inspiration or at end expiration. Two sets of pictures were taken at each lung volume and at each wavelength, for a total of 32 possible data sets.

Data at a transpulmonary pressure of 5 cmH2O were pooled with data at end expiration (low lung volume), and data at a transpulmonary pressure of 15 cmH2O were pooled with data at end inspiration (high lung volume).

Spectrophotometric procedures and dye calibration. To find f (Eq. 11) needed for our estimates of changes in blood volume, we mimicked known changes in blood volume or Hct by infusing ICG dye (Cardio-Green, Becton Dickinson) into a vein at constant rates (~0.375 and 0.75 mg/min) until the blood [ICG], as measured by serum OD, was constant. To avoid red cell lysis during ICG infusion, the dye was first dissolved in sterile water and then an approximately isosmotic solution was made by mixing the dissolved ICG with an equal volume of 1.8% saline. Blood samples were then taken from the arterial catheter, within 1 min after a series of pictures was taken (by using protocol II). Part of the sample was used for measurement of [Hb] and Hct (clinical laboratory, Brigham and Women's Hospital Boston, MA). The rest of each sample was stored upright in clean glass tubes for at least 2 h to allow the blood to clot and was then centrifuged for 12 min at 3,000 revolutions/min at room temperature (IECCentra-8R Centrifuge). The serum was then collected, and the OD was measured at 800 nm (bandwidth of 0.5 nm) (Gilford Response UV-VIS Spectrophotometer model 741).

Measurements were made in five animals before any ICG was infused, yielding OD < 0.1, and at the two different ICG infusion rates, yielding OD approx  1.5 and 3, respectively. All values of k2diff 800 were plotted against photometrically determined values of OD weighted by (1 - Hct) at end expiration and at end inspiration. The slope of these curves is an estimate of f (Eq. 11). We found a mean f of 2.15 ± 0.90 (SD) cm-1 at end inspiration and of 5.71 ± 1.85 cm-1 at end expiration. The origin of the interanimal variation is unknown.

Data analysis and statistics. We calculated delta Vc/Vc, <OVL>S</OVL>, and delta S in the region of lung below the light source from changes in the kdiff values by using Eqs. 12, 15, and 19. To examine the fluctuations in delta Vc/Vc and delta S over the cardiac cycle, we fit the measured values of kdiff to a sine wave in time, with a period equal to that of the cardiac cycle. The difference in the recorded time of the laser trigger and peak systolic BP was taken as time in the cardiac cycle, and the ratio of that interval to the BP peak-to-peak time of that cycle was taken as the fractional time in the cardiac cycle4. Denote this fractional time x. We fit observed values of kdiff and x, by the method of least squares, to the function
<IT>k</IT><SUB>diff</SUB>(<IT>x</IT>) = <OVL><IT>k</IT></OVL><SUB>diff</SUB> + <IT>a</IT> cos 2&pgr;<IT>x</IT> + <IT>b</IT> sin 2&pgr;<IT>x</IT> (20)
where <OVL><IT>k</IT></OVL>diff is the mean value over the cardiac cycle, and a and b are the cosine and sine components, respectively. This representation can also be written
<IT>k</IT><SUB>diff</SUB> = <OVL><IT>k</IT></OVL><SUB>diff</SUB> + <IT>M</IT> cos (2&pgr;<IT>x</IT> + &phgr;) (21)
where M is the magnitude (amplitude) of the variation (1/2 peak to peak) in kdiff secondary to the cardiac cycle, and phi  is the phase shift of these fluctuations relative to peak systemic BP. In this formulation phi  is the time lag of the fluctuations in kdiff, normalized to the cardiac cycle period.

For the data from protocol I (involving a short breath hold), we additionally used a "detrending" function (a linear change in kdiff with time), thus eliminating the progressive change in kdiff vs. time during the short breath hold, which was not related to the cardiac cycle.

To determine whether there were significant variations in delta Vc/Vc and delta S phasic with the cardiac cycle, we examined the F-statistic for analysis of variance (ANOVA) of kdiff with the sinusoidal fit. If the majority of the picture sets under a given set of conditions (e.g., high or low lung volume) resulted in a P < 0.05, then that kdiff was taken to vary phasically with the cardiac cycle. The rationale for this is explained in RESULTS AND DISCUSSION. delta Vc/Vc was taken to vary phasically with the cardiac cycle if kdiff 800 showed significant phasic variation, because delta Vc/Vc is estimated using only 800-nm light (delta Vc/Vc in Eq. 12). By contrast, the calculation of delta S requires measurements of kdiff at both 650 and 800 nm (Eq. 19), but because 800 nm is an isobestic wavelength, kdiff 800 is not sensitive to changes in S. Therefore, delta S was taken to vary phasically with the cardiac cycle if kdiff 650 showed significant phasic variation. We calculated the mean phase shift of delta Vc/Vc and delta S with the cardiac cycle by using the mean sine component and the mean cosine component.

To determine whether the average magnitudes of delta Vc/Vc and delta S over all animals were statistically significant, we calculated a Z statistic given by the root mean square (rms) magnitude over all animals divided by the SE. We tested whether delta Vc/Vc or delta S was significantly different from zero from the Z statistic corresponding to P < 0.05 (Z > 2).

To determine whether <OVL>S</OVL> was greater at end inspiration compared with end expiration, we performed a one-tailed t-test of paired calculated <OVL>S</OVL> values for the seven animals in which we had these data at both volumes. Average <OVL>S</OVL> values over repeat trials at each volume were used for each animal. <OVL>S</OVL> at end inspiration was significantly greater (P < 0.05) than at end expiration.

RESULTS AND DISCUSSION

Here we sequentially present results and discussion of measurements of <OVL>S</OVL>, delta Vc/Vc, and delta S, followed by a discussion of methodological issues and the potential importance of the DLS technique in further applications. Data were collected from eight animals. At low lung volume, acceptable measurements were made in 24 data sets at 650-nm wavelength and 23 data sets at 800-nm wavelength. At high lung volume, acceptable measurements were made in 23 data sets at 650 nm and 18 data sets at 800 nm. Data sets were excluded from analysis (n = 8) if there was inadequate pulse pressure in the systemic BP recording to allow accurate timing of the peak BP or if the pictures in a set sampled only one or two time points in the cardiac cycle.

<OVL><IT>S</IT></OVL>. <OVL>S</OVL> (averaged over the cardiac cycle and blood volume weighted), calculated from mean kdiff 650 and kdiff 800 by using Eq. 15, was 82 ± 3.2% (mean ± SE) at high lung volume and 77 ± 5.3% at low lung volume (P < 0.05). (High lung volume here refers to data collected at a transpulmonary pressure of 15 cmH2O pooled with data at end inspiration; low lung volume refers to data taken at a pressure of 5 cmH2O and at end expiration.) Note that nothing in the derivation of Eq. 15 constrains its result to have any particular value, in particular to values of 50-100%. Therefore, the finding that all results are well within the expected physiological range and are greater at end inspiration supports the validity of our theory and measurements. Given the shape of the oxyhemoglobin2 dissociation curve, a change in capillary saturation of 5% would be associated with a change in capillary PO2 of ~5 Torr. However, if we include the Bohr effect, we would expect a somewhat smaller change. This change in mean capillary PO2 (not end capillary) is consistent with tidal alveolar PO2 changes of 3-4.5 Torr, which have been predicted in humans (8, 12). With the rabbit's larger oxygen consumption-to-functional residual capacity (FRC) ratio, even larger fluctuations in alveolar PO2 would be expected.

An increase in <OVL>S</OVL> at high lung volume compared with low lung volume could be due on the one hand to 1) an increase in the kinetics of oxygen transport at the alveolar level secondary to either an increase in alveolar PO2 or an increase in septal surface area or, on the other hand, to 2) a change in the distribution of blood volume such that a greater proportion of the blood in our "volume of view" is postcapillary. It is certainly true that alveolar PO2 and septal surface area increase with lung volume and must, therefore, contribute to any increased mean microcirculatory saturation. On the other hand, it is currently unknown whether this mechanism suffices to explain the quantitative changes in <OVL>S</OVL> that we found or whether other mechanisms such as blood volume distribution are also important.

To assess the sensitivity of our optical measurements to changes in saturation, we examined changes in kdiff 650 in two animals during a breath hold of ~1-min duration at a continuous positive airway presure of 5 cmH2O and compared the results with simultaneously measured changes in SpO2 in the same animals. In the rabbit, FRC is small relative to oxygen consumption, so S falls precipitously during such a breath hold. Recall that as S falls absorption of 650-nm light increases, and, therefore, kdiff 650 should increase. As SpO2 fell, there was a striking increase in the optical absorption (Fig. 1). The quantitative results here must be viewed as only preliminary, however, for several reasons. First, SpO2 measurements are not accurate below saturations of ~70% (15, 18). More importantly, our probe measures the optical absorption of all blood in our volume of view and thus represents a mixture of venous, capillary, and arterialized blood. However, the vast majority of this is expected to be in the microcirculation (4). Nonetheless, to the extent that this average of mixed venous, capillary, and arterial saturation falls in a parallel fashion with SpO2 during a breath hold, the rate of change of kdiff with SpO2 should be comparable to that obtained from Eq. 15. These slopes differed by ~25%, which is highly suggestive that our methods are essentially correct. Furthermore, these data clearly show that kdiff 650 is a sensitive indicator of changes in S. Changes in kdiff 800, which would reflect changes in capillary blood volume only, made during the same type of breath hold were inconsistent in direction.
Fig. 1. Change in systemic arterial oxygen saturation measured with pulse oximeter vs. change in diffuse optical extinction coefficient for 650-nm light (kdiff 650) during ~1-min-long breath hold with transpulmonary pressure of 5 cmH2O. As systemic arterial oxygen saturation fell, there was a striking increase in optical absorption, demonstrating that kdiff 650 is a sensitive indicator of changes in saturation.
[View Larger Version of this Image (14K GIF file)]

Changes in kdiff values during the cardiac cycle. We found small but significant variations, phasic with the cardiac cycle, in kdiff 650 at both low and high lung volume (58 and 57% of data sets, respectively) and in kdiff 800 at low but not high lung volume (52 and 6% of data sets, respectively). There were no conditions for which kdiff 800 was significant and kdiff 650 was not. This is consistent with the fact that kdiff 650 is sensitive to both Vc and S, whereas kdiff 800 is sensitive only to Vc.

Our significance criterion (majority of picture sets with P < 0.05 at 800 and 650 nm for delta Vc/Vc and delta S, respectively) is a rough estimate of significance, because the F-test in ANOVA assumes a chi 2 distribution of variances. We do not know the distribution of variances in our fitting of kdiff, but they are certainly far from that required by a simple ANOVA. In particular, the significance of any given data set is strongly dependent on an adequate spread of the data over time throughout the cardiac cycle; i.e., in our experiments, the degree of asynchrony of the laser pulses and the heart rate. If only a few time points in the cardiac cycle are sampled, the estimate of a best-fit sine wave to those points is severely compromised. This effect is the likely origin of the observation that 50-60% of the individual data sets satisfied P < 0.05 rather than ~95%. On the other hand, the extremely low P values (median P < 0.0013) that we found for the majority do indicate that a significant variation in kdiff exists with the cardiac cycle, and we have, therefore, pooled all experiments (other than high lung volume-800 nm) into their respective three categories: high lung volume-650-nm, low lung volume-650-nm, and low lung volume-800-nm sets.

Figure 2 shows examples of the measured changes in kdiff 650 and kdiff 800 during the cardiac cycle. Figure 2 displays raw data points in a typical animal at low lung volume and the best sinusoidal fits (dotted lines) from Eq. 20 to the optical data. Time 0 corresponds to systemic systole. For technical reasons we were unable to time our measurements relative to changes in pulmonary arterial, rather than systemic, pressure.
Fig. 2. Examples of measured changes in diffuse optical extinction coefficient at 800 nm (kdiff 800; A) and 650 nm (kdiff 650; B) during cardiac cycle. Both panels display data in a typical animal at low lung volume, and time 0 corresponds to systemic systole. A: raw optical data points at 800 nm, best sinusoidal fit (dotted line) from Eq. 20, and translation of fit into sinusoidal fractional variation in capillary blood volume (delta Vc/Vc) during cardiac cycle (solid line, right-hand scale) by using Eq. 12. B: raw optical data points at 650 nm, best sinusoidal fit (dotted line) from Eq. 20, and translation of fits at both 650 and 800 nm into estimated sinusoidal variation in oxygen saturation delta S (solid line, right-hand scale) during the cardiac cycle by using Eq. 19. In this example, peak fractional variation in capillary blood volume lagged behind peak systemic blood pressure by ~<FR><NU>1</NU><DE>3</DE></FR> of a cardiac cycle, and change in microcirculatory oxygen saturation lagged the change in capillary blood volume by about another <FR><NU>1</NU><DE>3</DE></FR> of a cardiac cycle.
[View Larger Version of this Image (22K GIF file)]

delta Vc/Vc during the cardiac cycle. Figure 2A shows an example of raw data and the best sinusoidal fit at 800 nm (dotted line) and the translation of the fit into the sinusoidal variation in delta Vc/Vc during the cardiac cycle (solid line, right-hand scale) by using Eq. 12, in a typical animal at low lung volume. In this example the peak delta Vc/Vc lagged behind the peak systemic BP by about one-third of a cardiac cycle. The arithmetic mean ± SE of the variation in delta Vc/Vc during the cardiac cycle in eight animals, at low lung volume, is shown as a polar plot in Fig. 3. The magnitude of the average variation in delta Vc/Vc was 3.9%; the rms variation was 6.4% (P < 0.05) (a peak-to-peak variability of ~13%). The phase of delta Vc/Vc lagged systemic systole by ~64° or 0.18 of a cardiac cycle (~0.053 s for a heart rate of 200 beats/min). By contrast, at high lung volumes (6 animals), there was no significant variation in delta Vc/Vc.
Fig. 3. Polar plot (360° is 1 cardiac cycle) of arithmetic mean ± SE of fractional variation in Vc (delta Vc/Vc) and variation in microcirculatory oxygen saturation (delta S) in 8 animals at low lung volume. Phase lags of delta Vc/Vc and delta S relative to peak systemic blood pressure are given by their respective clockwise angles from positive x-axis (systole). Difference in timing of delta S and delta Vc/Vc is consistent with the fact that right ventricular systole increases Vc secondary to an influx of venous blood (with a low oxygen saturation), which is then oxygenated, resulting in an increasing S. We believe the timing of the increase in S is largely due to oxygen-loading kinetics.
[View Larger Version of this Image (9K GIF file)]

The 13% variation we found in blood volume, at end expiration during the cardiac cycle, suggests that not only are intravascular pressure fluctuations transmitted from the right ventricle to the level of the pulmonary capillary but also that the capillary volume changes in response to these pressure variations. The size of the variation is close to the upper bound of ±10% found by Fowler and Maloney (6), who used an external gamma counter after injection of labeled red blood cells. By contrast, Menkes et al. (9) found changes in capillary volumes manyfold greater than what we found. On the other hand, they also recognized that their results were completely unphysiological, with implied variations so large as to require not only retrograde flow from the venous side of the circulation but also in amounts larger than the left atrial stroke volume. We know of no other measures of fluctuation in the pulmonary capillary volume during the cardiac cycle, although others have modeled the changes in dogs (5) and changes in flow have been demonstrated (14).

Our findings of significant fractional change in blood volume density during the cardiac cycle at low lung volume but not at high lung volume could be due to real differences in delta Vc/Vc during the cardiac cycle or to a greater absolute Vc at high lung volume with the same stroke volume, so delta Vc/Vc would be smaller with the same delta Vc. Although an increase in blood volume density at high lung volume is possible, we expect no change or a decrease in mean Vc as lung volume increases. The most likely causes of tidal changes in delta Vc/Vc are an end-inspiratory decrease in capillary compliance (due to increased tissue forces tending to stiffen the vasculature) or a decrease in the pressure fluctuations to which the capillaries are exposed (due to decreased right heart stroke volume or because the pressure fluctuations become damped outside our volume of view, i.e., in the pulmonary artery). Further experiments will be required to determine the extent to which capillary compliance varies with lung volume.

If pulmonary capillaries had the 1-3%/mmHg distensibility of 30- to 70-µm pulmonary vessels described by others in dogs (1, 7), a 13% change in Vc would require a 4- to 13-mmHg change in pressure. On the basis of a capillary compliance of 4-10%/mmHg, derived from the rabbit data of Bachofen et al. (2), a 13% change in Vc would require a 1- to 3-mmHg change in capillary pressure. These values for pressure fluctuations in the pulmonary capillary are consistent with expected values.

Much has been written about the change in capillary flow during the cardiac cycle, but less is known about the changes in blood volume and compliance because they could not be measured in vivo. Vc is one of the fundamental parameters in determination of pulmonary gas exchange through its influence on pulmonary diffusing capacity (13). Capillary compliance and area are important determinants of the pattern of pulmonary perfusion. To the degree that capillary length changes with lung volume, but not during the cardiac cycle at fixed lung volume, a 13% change in Vc would be associated with a 13% change in cross-sectional area (or 6.3% change in diameter) if the volume change were due to distension and not recruitment. Changes in area of this size could have important effects on neutrophil transit times through the pulmonary vasculature.

delta S during the cardiac cycle. Figure 2B shows an example of the measured changes in kdiff 650 (points), the best sinusoidal fit (dotted line), and translation of the fits at 650 and 800 nm into the estimated sinusoidal variation in delta S (solid line, right-hand scale) during the cardiac cycle from Eq. 19, in a typical animal at low lung volume. In this example, delta S lagged behind the change in capillary blood volume by about one-third of a cardiac cycle and lagged behind the peak systemic BP by about two-thirds of a cardiac cycle. The arithmetic mean ± SE of the variation in delta S during the cardiac cycle in eight animals at low lung volume is shown as a polar plot in Fig. 3. The magnitude of the average variation in delta S, calculated with Eq. 19, was 1.2%; the rms variation was 2.3% (P < 0.05) (a peak-to-peak variability of ~2.6%). The phase of delta S lags delta Vc/Vc by ~146° or 0.41 of a cardiac cycle (~0.122 s for a heart rate of 200 beats/min). The difference in timing of delta S and delta Vc/Vc is consistent with the fact that right ventricular systole increases Vc secondary to an influx of venous blood (with a low oxygen saturation), which is then oxygenated, resulting in an increasing S. We believe the timing of the increase in S is largely due to oxygen-loading kinetics. At high lung volumes (6 animals) the rms variation in S was <1%, and although statistically significant, it was physiologically unimportant.

Methodological issues. All our measurements of fractional change in blood volume density and saturation are blood volume-weighted means of all blood in our volume of view. This volume is roughly hemispherical with a diameter of 1-2 cm and, therefore, penetrates well into the lung. Evidence for the depth of penetration is twofold: 1) the pattern of light we observe at the pleural surface is consistent with the theoretical pattern of light diffusively scattered in a semi-infinite isotropic homogeneous medium; in such cases the depth of penetration is approximately equal to the lateral spread of light seen at the surface, and 2) we previously found similar depths of penetration of light through tapering lung segments.

We believe the greatest source of potential error and interanimal variability in our measurements (of S and delta S and the phase of delta S relative to BP and delta Vc/Vc) relates to the fact that we made measurements by using 650- and 800-nm light at very different times because only a single laser was available. However, this limitation can easily be overcome with simultaneous measurements at two wavelengths by using multiple light sources and wavelength-sensitive methods to characterize the pattern of backscattered light. Another limitation of our methodology is our inability to precisely quantitate the Anb. However, we believe the changes in absorption due to nonblood elements are small on the basis of previous measurements made by our laboratory (17), in which 632-nm light was used. Further variability in our measurements may have been caused by imprecise control of lung volume history and of lung volume itself, especially at end inspiration when the laser was triggered off the ventilator cam.

DLS technology. The standard measures of A/V values, as well as capillary compliances, rest on stereologic techniques, wherein fixed specimens are examined by microscopic morphometry. The greatest strength of this technique is that, while labor intensive, it is a reproducible technique that, depending on the tissue sample size, can yield results of essentially arbitrarily fine precision. By contrast, its greatest weakness is that, by definition, stereology of fixed specimens cannot even in principle address either static issues in the living animal or dynamic issues at all. The use of DLS is, to our knowledge, the only available means to even estimate the magnitude of, e.g., fractional changes in these variables. In this light, we now have a unique tool by which to address both static and dynamic questions in living animals, with a wide variety of potential interventions, including perturbations and disease models of the lung, as well as lung function changes secondary to changes in right ventricular and left atrial pressures.

Conclusions. We have shown that <OVL>S</OVL>, delta S, and delta Vc/Vc can be measured in living animals by observing the pattern of diffusely scattered light at multiple wavelengths. At low, but not high, lung volumes there were physiologically important fluctuations in mean capillary saturation (4.6%) and capillary blood volume (13%) during the cardiac cycle. In addition, the mean capillary saturation varied with the respiratory cycle, from 77% at low lung volume to 82% at high lung volume. These results are consistent with fluctuations in gas exchange and pulmonary capillary intravascular pressure over the cardiac cycle and with decreasing capillary compliance with increasing transpulmonary pressure. These values, obtained in normal animals, should serve as a baseline to be compared with other normal physiological states (e.g., exercise), various pathophysiological states (e.g., adult respiratory distress syndrome, pulmonary infection, pulmonary hypertension), and in the presence of various drugs that are thought to have profound effects on the pulmonary vasculature (e.g., varying inspired oxygen fraction, histamine, nitric oxide, volatile anesthetics). This technique has promise not only as a research tool but also as a guide to clinical strategies such as intraoperative mapping of lung structure and function as an aid in lung reduction surgery.

ACKNOWLEDGEMENTS

This work was supported in part by National Heart, Lung, and Blood Institute Grants R43 HL-55032 and PO1 HL-33009.

FOOTNOTES

1   For simplicity, we will use the term scattering to refer to refraction and reflection, although strictly speaking absorption is also a form of scattering.

2   epsilon HbO2 650 = 0.095, epsilon Hb 650 = 0.855, and epsilon HbO2 800 = epsilon Hb 800 = 0.22 cm-1 · l · mmol-1, where the subscript includes the species name and the wavelength.

3   Although reported values for the isobestic wavelength vary from 800 to 815 nm, the differences in extinction coefficients over this range are small (16, 19, 20, 22).

4   For technical reasons in one animal in protocol I, the time was measured relative to the peak in a sine-wave model of the BP signal.

Address for reprint requests: G. P. Topulos, Dept. of Anesthesia, Brigham and Women's Hospital, 75 Francis St., Boston, MA 02115 (E-mail: topulos{at}zeus.bwh.harvard.edu).

Received 16 January 1996; accepted in final form 26 December 1996.

REFERENCES

1. Al-Tinawi, A., A. Clough, D. Harder, J. Linehan, D. Rickaby, and C. Dawson. Distensibility of small veins of the dog lung. J. Appl. Physiol. 73: 2158-2165, 1992 [Abstract/Free Full Text] .
2. Bachofen, H., D. Wangensteen, and E. Weibel. Surfaces and volumes of alveolar tissue under zone II and zone III conditions. J. Appl. Physiol. 53: 879-885, 1982. [Abstract/Free Full Text]
3. Butler, J., H. Miki, S. Suzuki, and T. Takishima. Step response of lung surface-to-volume ratio by light-scattering stereology. J. Appl. Physiol. 67: 1873-1880, 1989 [Abstract/Free Full Text] .
4. Doerschuk, C., J. Markos, H. Coxson, D. English, and J. Hogg. Quantitation of neutrophil migration in acute bacterial pneumonia in rabbits. J. Appl. Physiol. 77: 2593-2599, 1994 [Abstract/Free Full Text] .
5. Fishman, A. Pulmonary circulation. In: Handbook of Physiology. The Respiratory System. Circulation and Nonrespiratory Functions. Bethesda, MD: Am. Physiol. Soc., 1985, sect. 3, vol. I, chapt. 3, p. 93-165.
6. Fowler, K., and J. Maloney. A search for a pulsatile component in pulmonary capillary blood volume. J. Appl. Physiol. 20: 1173-1178, 1965. [Abstract/Free Full Text]
7. Hillier, S., P. Godbey, C. Hanger, J. Graham, R. Presson, O. Okada, J. Linehan, C. Dawson, and W. Wagner. Direct measurement of pulmonary microvascular distensibility. J. Appl. Physiol. 75: 2106-2111, 1993 [Abstract/Free Full Text] .
8. Hlastala, M. A model of fluctuating alveolar gas exchange during the respiratory cycle. Respir. Physiol. 15: 214-232, 1972 [Medline] .
9. Menkes, H., K. Sera, R. Rogers, R. Hyde, R. Forster II, and A. DuBois. Pulsatile uptake of CO in the human lung. J. Clin. Invest. 49: 335-345, 1970 .
10. Miki, H., J. Butler, R. Rogers, and J. Lehr. Geometric hysteresis in pulmonary surface-to-volume ratio during tidal breathing. J. Appl. Physiol. 75: 1630-1636, 1993 [Abstract/Free Full Text] .
11. Morse, P., and H. Feshbach. Equations governing fields. In: Methods of Theoretical Physics. New York: McGraw-Hill, 1953, pt. 1, chapt. 2, p. 171-200.
12. Otis, A. Quantitative relationships in steady-state gas exchange. In: Handbook of Physiology. Respiration. Washington, DC: Am. Physiol. Soc., 1964, sect. 3, vol. I, chapt. 27, p. 681-698.
13. Roughton, F., and R. Forster. Relative importance of diffusion and chemical reaction rates in determining rate of exchange to true diffusing capacity of pulmonary membrane and volume of blood in the lung capillaries. J. Appl. Physiol. 11: 290-302, 1957. [Abstract/Free Full Text]
14. Sackner, M. Measurement of cardiac output by alveolar gas exchange. In: Handbook of Physiology. The Respiratory System. Gas Exchange. Bethesda, MD: Am. Physiol. Soc., 1987, sect. 3, vol. IV, chapt. 13, p. 233-255.
15. Severinghaus, J., and S. Koh. Effect of anemia on pulse oximeter accuracy at low saturation. J. Clin. Monit. 6: 85-88, 1990 [Medline] .
16. Sevick, E., B. Chance, J. Leigh, S. Nioka, and M. Maris. Quantitation of time-and-frequency-resolved optical spectra for the determination of tissue oxygenation. Anal. Biochem. 195: 330-351, 1991 [Medline] .
17. Suzuki, S., J. Butler, E. Oldmixon, and F. Hoppin. Light scattering by lungs correlates with stereological measurements. J. Appl. Physiol. 58: 97-104, 1985 [Abstract/Free Full Text] .
18. Tremper, K., and S. Barker. Pulse oximetry. Anesthesiology 70: 98-108, 1989 [Medline] .
19. Tripp, M., C. Swayze, and I. Fox. Indocyanine green. In: Dye Curves: The Theory and Practice of Indicator Dilution, edited by D. A. Bloomfield. Baltimore, MD: University Park, 1974, p. 365-391.
20. Van Assendelft, O. Spectrophotometry of Haemoglobin Derivatives. Springfield, IL: Thomas, 1970.
21. Weibel, E. Stereological methods. In: Practical Methods for Biological Morphometry. New York: Academic, 1979, vol. 1, p. 36-37.
22. Zijlstra, W., and G. Mook. Dye dilution methods. In: Medical Reflection Photometry. Assen, The Netherlands: Van Gorcum, 1962, chapt. IX, p. 168-222.
0161-7567/97 $5.00 Copyright © 1997 the American Physiological Society