relationship of ultrafiltration and anastomotic flow in isolated rat lungs

Relationship of Ultrafiltration andAnastomotic Flow in Isolated Rat Lungs

WEN LIN, GENEVIEVE HOGAN, AND RICHARD M. EFFROSDivision of Pulmonary and Critical Care Medicine, Department of Medicine,

Medical College of Wisconsin, Milwaukee, WI, USA; and Zablocki V.A.Medical Center, Milwaukee, WI, USA

ABSTRACT

Objective: When arterial and venous pressures are increased to equal values in“stop-flow” studies, perfusate continues to enter the pulmonary vasculaturefrom the arterial and venous reservoirs. Losses of fluid from the pulmonaryvasculature are due to ultrafiltration and flow through disrupted anastomotic(bronchial) vessels. This study compared the relative sites of ultrafiltration andanastomotic flows at low and high intravascular pressures.Methods: Isolated rat lungs were perfused for 10 minutes with FITC-dextran,which was used to detect ultrafiltration. Arterial and venous catheters were thenconnected to reservoirs containing radioactively labeled dextrans at 20 or 30 cmH2O for 10 minutes. The vasculature was subsequently flushed into serial vials,and ultrafiltration and vascular filling during the equal-pressure interval werecalculated.Results: Ultrafiltration equaled 0.43 ± 0.11 mL at 20 cm H2O and was similarto the volume of fresh arterial and venous perfusate which entered and remainedin the pulmonary vasculature during the equal-pressure interval (0.45 ± 0.10mL). At 30 cm H2O, 0.80 ± 0.23 mL entered and remained in the vasculatureduring the equal-pressure interval, replacing the original perfusate, and calcu-lated transudation (0.56 ± 0.09 mL) was not significantly more than at 20 cmH2O. Fluid also entered the airspaces at 30 cm H2O but not at 20 cm H2O.Conclusions: At 20 cm H2O, flow through anastomotic vessels occurs at sitesthat are at the arterial and venous ends of the microcirculation. Flow in exchangevessels remains minimal, permitting measurements of ultrafiltration and ex-change. Losses of perfusate from the pulmonary vessels complicate measure-ments of ultrafiltration at 30 cm H2O. Microcirculation (2001) 8, 321–334.

KEY WORDS: bronchial vessels, pulmonary edema, pulmonary vasculature, tran-sudation

INTRODUCTION

Most studies of hydrostatic edema formation in iso-lated lungs have been conducted by measuring in-creases in organ weight after increasing intravascu-lar pressure. It has been assumed that when leftatrial pressures are raised, early increases in lungweight reflect vascular distention and recruitment,whereas late increases are due to ultrafiltration(transudation) of fluid into the tissues (4). This as-

sumption may be inaccurate because evidence hasbeen obtained that vascular distention can occurover a period of many minutes (2,7,10). Further-more, it is possible that compliant regions of theinterstitium fill rapidly soon after intravascular pres-sures are increased. These considerations have led toa variety of indicator dilution approaches to detectlosses of protein-free fluid from the vasculature(2,8,14,15). Increases in pulmonary venous concen-trations of an appropriate macromolecule over con-centrations entering the pulmonary artery should re-flect ultrafiltration. Unfortunately, losses of fluidfrom the vasculature are very small compared to in-travascular flows, making it difficult to detect ultra-filtration during single-pass studies. The perfusatecan be recirculated so that cumulative increases in

Supported by National Institutes of Health Grant No. HL 60057.For reprints of this article, contact Richard M. Effros, M.D., 9200West Wisconsin Ave., Milwaukee, WI 53226, USA; e-mail:[email protected] 3 January 2001; accepted 22 April 2001

Microcirculation (2001) 8, 321–334© 2001 Nature Publishing Group 1073-9688/01 $17.00www.nature.com/mn

concentrations can be followed, but this approachmakes it difficult to assess early transudation andmay expose the arterial end of the vasculature tometabolites released from vessels at the venous end.

As an alternative to lung weight, single-pass, andrecirculating indicator dilution approaches, we haveinvestigated ultrafiltration and solute exchange withstop-flow studies in isolated rat lungs (6,9). Theseexperiments are analogous to stop-flow experimentsin the kidney (11). Lungs are perfused with a solu-tion that contains a macromolecular (“intravascu-lar”) indicator. The perfusion pump is then turnedoff and the venous and arterial catheters are con-nected to reservoirs at equal heights above the lungs.Following a 10-minute interval, the fluid remainingin the lungs is flushed from the lungs. Increases inintravascular concentrations of the intravascularmacromolecule can be used to measure the amountof ultrafiltration that occurred during the equal-pressure interval. In principle, this approach has twoadvantages: 1) the volume of fluid from which ul-trafiltration occurs is limited to that within the vas-culature, thereby magnifying increases in macromo-lecular concentration, and 2) comparisons of earlyand later samples collected from the outflow canprovide information concerning the site of ultrafil-tration in the pulmonary vasculature. Using this ap-proach, we obtained information that suggests thatmore ultrafiltration occurs at the venous end of ex-change vessels than at more arterial sites (9).

Ideally, flow should eventually cease when the arte-rial and venous vessels are exposed to equal hydro-static pressures. However, perfusate continues to en-ter the pulmonary vasculature for at least 10 min-utes after the arterial and venous reservoirs are set atequal heights above the lungs. Because flow neverceases completely during these studies, it is moreappropriate to refer to the interval during which ar-terial and venous pressures are equalized as the“equal-pressure” rather than the “stop” interval,and we will use this terminology in this report. Per-sistent flow into the pulmonary vasculature can beattributed to several factors: 1) ultrafiltration, 2)distention and recruitment of pulmonary vessels,and 3) losses from the vasculature through disruptedvessels. Tears may occur in either the pulmonary oranastomotic (e.g., bronchial) vessels, allowing per-fusate to escape into the tissues, the airspaces, andenvironment.

Reliable measurements of ultrafiltration and ex-change with equal-pressure-flow experiments canonly be made if most of the perfusate that was origi-

nally within the pulmonary exchange vessels re-mains there throughout the equal-pressure interval.If losses of perfusate are restricted to precapillaryand postcapillary vessels and pressures are equal atthe arterial and venous ends of the microcirculation,flow should be minimal in the exchange vessels ofthe lungs. The objective of this study was to deter-mine if flow in the pulmonary exchange vessels iseffectively arrested during equal-pressure intervalwhen the pulmonary arterial and venous pressureswere equalized at 20 or 30 cm H2O. This wouldindicate that the fluid entering the lungs during theequal-pressure interval flowed out of bronchial ves-sels at the arterial and venous ends of the pulmonaryexchange vessels.

MATERIALS AND METHODS

General Approach

The pulmonary vasculature was initially loaded witha macromolecular indicator (FITC-dextran, mol. wt.2000 KDa) [see Table 1 and Fig. 1(A)], and thenflow was discontinued for 10 minutes. During thisequal-pressure interval, the pulmonary artery andvein were connected to arterial and venous reservoirsthat were set at either 20 cm or 30 cm above thelungs. Thereafter, the pulmonary vasculature wasflushed with perfusate and fluid was collected fromthe venous outflow. Increases in concentrations offluorescein isothiocyanate (FITC)-dextran were usedto monitor ultrafiltration from the pulmonary vas-culature which had occurred during the equal-pressure-flow interval. Radioactively labeled macro-molecular indicators were placed in the arterial andvenous reservoirs. This permitted calculation of thefraction of the samples collected at the end of theequal-pressure interval which entered the vascula-ture during the equal-pressure-flow interval fromthe arterial and venous reservoirs. At the end of theequal-pressure-flow interval, the lungs were flushedwith a solution containing another macromolecule(indocyanine green, ICG, a dye that binds to albu-min) (5,7) so that the total vascular volume could bedetermined. Finally, the airways were flushed with afifth macromolecule (rhodamine-labeled dextran) topermit detection of fluid that had entered the air-spaces during the experiment. The experimental ap-paratus is shown in Figs. 1(B) and 1(C).

Surgery

Sprague-Dawley rats, with an average weight of 328± 37 grams (SD, weighed in nine of the ten studies),were used in these studies. The rats were anesthe-tized with an intraperitoneal injection of 50 mg/kg

Pulmonary transudation and anastomotic flowW Lin et al.

322

pentobarbital. The chest was opened, P.E. 240 cath-eters were inserted into the pulmonary artery andthe common pulmonary vein, a P.E. 200 catheterwas placed in the trachea, and the lungs were ex-cised. The lungs were inflated with several breaths of3 mL of air to minimize atelectasis, and then keptinflated at a constant pressure of 5 cm H2O. Theywere then perfused at 6 mL/min. The perfusion so-lution contained 5 g/L bovine serum albumin (Bo-vine, Fraction V; Sigma, St. Louis, MO), 130 mEq/Lsodium, 4 mEq/L potassium, 2.7 mEq/L calcium,109 mEq/L chloride, 28 mEq/L lactate, 25 mMHEPES buffer (N-[2-hydroxethylpiperazine-N8-2-ethanesulfonic acid], Sigma, St. Louis, MO), 1 mMmagnesium sulfate, and small volumes of 1N HCl toderive a solution with a pH of 7.4. All of the perfu-sion solutions contained the same concentration ofFITC-dextran (see Table 1).Load Interval (10 Minutes)

The lungs were mounted in a 37 °C jacketed cham-ber. They were initially perfused from reservoir I,

which contained only one macromolecule (FITC-dextran) (see Fig. 1). During the loading period,flow was kept at 6 mL/min and left venous pressurewas kept at 6 cm H2O (1 cm above airway pressure).Pressures in the pulmonary artery and vein weremonitored throughout the experiment with an ana-log-to-digital interface and a computerized spread-sheet program.

Samples were collected from the outflow with a Gil-son Fraction Collector synchronized by a dropcounter so that samples were collected at approxi-mately 2-second intervals. Collection began 20 sec-onds before the equal-pressure interval was com-menced and tubes were changed every 3 drops. Thisvirtually prevented losses of fluid between samples.The volume of fluid in each tube was determined byweighing the tubes. The time of collection for eachtube was also monitored and recorded electronically,permitting continuous calculation of the flow of fluidemerging from the pulmonary vein.

Equal-Pressure Interval (10 Minutes)

After the loading period was completed, the arterialand venous catheters were connected to reservoirssituated at equal distances (20 or 30 cm) above thelungs. To ensure that the height of the perfusate inthe reservoirs remained unchanged throughout theequal-pressure interval, fluid was continuouslypumped from a storage cylinder (S.C.) into the topof the reservoirs and allowed to spill over into anouter tube (O.T.), as shown in Fig. 1(C). The over-flow was then returned to a storage cylinder. By ad-justing the pumps, it was possible to keep the top ofthe reservoirs at constant and equal levels. Losses offluid from the arterial and venous cylinders weremonitored during the equal-pressure intervals bymeasuring the hydrostatic pressure (Pres) in a poly-ethylene tube that was immersed in the storage cyl-inder.14C-dextran and 3H-dextran were added to eitherthe arterial or venous perfusate solutions so that thepresence of fluid from these reservoirs could be de-tected in the collection tubes after the equal-pres-sure-flow interval was completed (see below).

Flush Interval (5 Minutes)

At the end of the equal-pressure interval, the pul-monary vasculature was flushed at 2 mL/min withsolution from reservoir II (Fig. 1). Flow was kept lowto minimize further transudation during the flushinterval. Indocyanine green was incorporated in theflush solution so that the vascular volume at the end

Table 1. Macromolecular indicators (see Fig. 1)

IndicatorInitial

location Function

FITC-dextran(BD) (mol. wt.2000 KDa)(Sigma) (100mg/L)

All reservoirs(not inairwaysolution)

Measure transudationfrom vasculature

3H-dextran (ARC)(mol. wt. 70KDa) (10mCi/L)

Arterial orvenousreservoirs

Determine fraction ofoutflow samplesderived fromarterial or venousreservoirs duringthe equal-pressure-flow interval

14C-dextran(ARC) (mol. wt.70 KDa) (10mCi/L)

Arterial orvenousreservoirs

Determine fraction ofoutflow samplesderived fromarterial or venousreservoirs duringthe equal-pressure-flow interval

Indocyanine green(2.5 mg/L)(BD)

Reservoir II Determine thevascular volumeduring the flushinterval

Rhodamine-dextran(mol. wt. 70KDa) (Sigma)(0.5 mg/L)

Flushed intothe airspacesat the end ofthe study

Detect entry of fluidinto the airspaces

Note: ARC 4 American Radiolabeled Chemicals (St. Louis, MO); BD4 Becton Dickenson (Cockeysville, MD); Sigma (St. Louis, MO).


323

of the equal-pressure-flow interval could be calcu-lated (see below). Samples were collected from theoutflow with the timed drop collector as describedabove, and the volumes of fluid in each collectiontube were determined gravimetrically.

Lung Lavage

At the end of the flush interval, 5 mL of the perfus-ate labeled only with rhodamine-dextran (see Table1) was injected into the trachea. As much fluid ascould be recovered was then removed and reinjectedthree times to encourage mixing and was then placedin a sample tube. Dilution of the rhodamine-dextranwas used to determine the volume of fluid in theairspaces at the end of the experiments.

Sample Analysis

Samples were centrifuged for 10 minutes at 10,000G prior to these measurements to remove residual

red cells and other debris. Concentrations of dyeswere measured spectrophotometrically (FITC-dextran at 495 nm, ICG-albumin at 800 nm, andrhodamine-dextran at 555 nm). Concentrations of14C-dextran and 3H-dextran were determined byadding 0.1 mL of the outflow samples to 2 mL ofLisquiscint scintillation fluid (National Diagnostics,Atlanta, GA), and the scintillation vials werecounted in an automated beta counter. Correctionwas made for background counts and indicatorcross-overs.Volume and Recovery Calculations

The radioactive counts per minute or the opticaldensity of each indicator were divided by the com-parable values of the corresponding baseline solu-tions to yield fractional concentrations, C.

The volume of fluid (DVtrans) that was lost by tran-sudation from the fluid that entered the collection

Figure 1. Experimental procedure. Parts are not drawn to scale and 37 °C water jackets surrounding the four reservoirsand lungs are not shown. Section A indicates the sequence of events. “Load” designates the initial period in which thelungs are loaded with FITC-dextran. “Stop” indicates the interval during which arterial and venous pressures areequalized. “Flush” represents the period during which the vasculature is flushed with fluid containing ICG-albumin.The dead space of the apparatus is divided into arterial (art), venous (ven), and collection (col) components that areseparated by the connectors shown in section B. In section B, reservoir I contains FITC-dextran, which is used to loadthe lungs prior to the equal-pressure interval. Reservoir II contains FITC-dextran and ICG-albumin. It is used to flushthe vasculature at the end of the equal-pressure interval. The arterial and venous reservoirs are connected to thepulmonary catheters during the equal-pressure-flow interval. They contain 3H-dextran and 14C-dextran as well asFITC-dextran. The overflow feature (see section C of the figure) keeps the head of pressure in arterial and venousreservoirs constant. Fluid is pumped from the storage chamber (S.C.) into the reservoir (Res) and overflows into theoverflow tube (O.T.). It is pumped from the overflow tube back into the storage chamber. A fluid-filled catheter isplaced in the storage chamber and the pressure in this tube (Pres) is used to determine the volume of perfusate remainingin the storage chamber.


324

samples was calculated from a) the fractional con-centration of FITC-dextran emerging from the pul-monary vein prior to the equal-pressure intervals(CFITC,baseline), b) the fractional concentration ofFITC-dextran in the outflow sample (CFITC,sample),and c) the volume of the sample (DVsample), with theequation:

DVtrans

= SCFITC,sample − CFITC,baseline

CFITC,baselineD DVsample

(1)The total volume of fluid, Vtrans, ultrafiltered fromthe samples was calculated from the equation:

Vtrans = (i=l

i=n

DVtrans (2)

where n designates the last collection sample.

The volume, DVICG, of the perfusate labeled withICG-albumin (which was used to flush the pulmo-nary vasculature) present in each of the collectionsamples was calculated from the fractional concen-trations, CICG,sample, of ICG-albumin and the volumeof the samples with the equation:

DVICG = CICG,sampleDVsample (3)The perfusate that was present in the vasculature atthe end of the equal-pressure interval could not con-tain any ICG-albumin, because this indicator wasintroduced during the flush interval. The volume ofthe perfusate, Vvascular, which was in the vasculaturejust prior to the flush period was calculated with theequation:

Vvascular = (i=l

i=n

~1 − DVICG! + Vresidual

− ~Dart + Dven + Dcoll! (4)

where values of D represent the dead-space compo-nents shown in Fig. 1(A): Dart 4 0.66 mL, Dven 40.24 mL, and Dcol 4 1.20 mL (these dead-spacevolumes were measured volumetrically). Even at theend of the flush period, small volumes of residualperfusate that did not contain ICG-albumin re-mained in the samples, indicating that some of thefluid (designated Vresidual) that was in the pulmo-nary vasculature at the end of the equal-pressureinterval was not flushed out by the ICG-albuminsolution. To estimate the value of Vresidual, it wasassumed that the washout of the unlabeled indicatorwas monoexponential after the nth sample and couldbe calculated from the downslope of H 4 (1 − CICG)prior to the nth sample. Values of ln(1 − VICG) wereplotted against outflow volume (see Fig. 8) and the

least squares best fit slope of the final samples, k,was calculated. Vresidual was derived from:

dHdV

= −kH

H = Hne−kVn

Vresidual = *0

`

Hne−kVe dVe

= FHne−kVe

−k G0

`

=Hn

k=

l − CICG,n

k(5)

where Ve represents the volume of perfusate thatwould have been present in each of the outflowsamples had collection continued after the lastsample (assuming that the rate of perfusion was keptconstant). CICG,n designates the fractional concen-tration of ICG in the last collected sample. Fractionalconcentrations of the arterial and venous indicatorsin the outflow were corrected for changes due totransudation by multiplying them by CFITC,baseline/CFITC,sample.

Many of the samples collected during the flush pe-riod contained perfusate that was derived from ei-ther the arterial or venous reservoirs. The volume ofperfusate from the arterial or venous reservoirs(DVx, where x represents arterial or venous) presentin each sample was:

DVx = Cx,sampleDVsample (6)

where the fractional concentration, Cx,sample, desig-nates the concentration of x in the sample divided bythe concentration in the reservoir containing x.

The total volume, Vx, of the arterial or venous per-fusate in the pulmonary vasculature at the time ofthe flush was calculated from:

Vx = (i=l

i=n

DVx − Dx (7)

where Dx is Dart or Dven.

The volume of the original loading solution fromreservoir I (DVload) in each sample was calculatedfrom:

DVload = ~l − CICG − Cven − Cart! DVsample (8)


325

The total volume of the original loading solutionwas:

Vload = (i=l

i=n

DVload − Dcol (9)

In order to detect fluid that was in the airspaces atthe end of the experiments, 5 mL of perfusate la-beled with rhodamine-dextran was instilled into theairspaces. Concentrations of this dye remained un-changed in the 20 cm H2O experiments, but in the30 cm H2O experiments, rhodamine-dextran con-centrations were lower in lavage samples than in thefluid that was instilled into the airspaces. The de-crease in concentration at 30 cm H2O indicated thatthere was fluid already present in the airspaces. Thevolume of fluid, DVairspace, in the airspaces prior tolavage was calculated from the fractional concentra-tion of rhodamine instilled into the lungs (CR,lavage)and the fractional concentration in the collectedsamples (CR,airspace):

DVairspace = SCR,lavage − CR,airspace

CR,airspaceD Vlavage

(10)

where Vlavage was the volume instilled into the air-ways for lavage (5 mL).

Of the 14C- and 3H-labeled dextrans that were lostfrom the arterial and venous reservoirs, the fractionsthat were subsequently recovered in the venous out-flow were calculated by summing the amounts re-covered in the venous outflow, the airspace fluid,and the thermostated jacket and dividing the sumsby the amounts that were lost from the reservoirs.

Statistical Analysis

When two mean values were compared, a t-test wasused to test for significance of differences. Compari-sons of multiple means were accomplished by a one-way ANOVA with a Newman-Keul test to test indi-vidual mean differences (Sigmastat version 2; JandelCorp., San Rafael, CA).

RESULTS

The observations made during these experimentscan be divided into three intervals. During the load-ing interval, the lungs were perfused with a solutionthat contained FITC-dextran. This was followed byan equal pressure interval, during which time thearterial and venous pressures were equalized. In theflush interval, the fluid remaining in the vasculaturewas flushed out with fresh perfusate.

Hemodynamics

Loading Interval. The lungs were initially perfusedat 6 mL/min with a solution containing FITC-dextran. During this interval, arterial pressures av-eraged 12 ± 6 (S.E.M.) cm H2O in the 20 cm H2Oexperiments and 15 ± 6 cm H2O in the 30 cm H2Ostudies (these differences were not significant) (Fig.2). Venous pressures were kept at 5 to 6 cm H2O andairway pressures were kept at 4 cm H2O.

Equal Pressure Interval. As indicated in Fig. 2, ar-terial and venous pressures were effectively keptconstant and equal throughout the 20 and 30 cmH2O experiments.

Flush Interval. In order to avoid excessive pres-sures during the flush interval, flow was reduced to2 mL/min and left atrial pressures were reduced tozero (Fig. 2). Pulmonary arterial pressures fell to 4 ±1 cm H2O in the 20 cm H2O studies and to 6 ± 2 cmH2O in the 30 cm H2O studies. The rates at whichperfusate left the lungs during the loading and flushintervals are indicated in Fig. 3. It will be noted thatthere is a transient surge in flow rates after the ar-terial and venous catheters were disconnected fromthe reservoirs and flow was resumed.

Figure 2. Arterial and venous pressures. Note that theoverflow reservoir system shown in Fig. 1 keeps arterialand venous pressures virtually the same during the equal-pressure-flow intervals.


326

Loss of Perfusate from the Reservoirs during theEqual-Pressure Interval

Perfusate continued to enter the pulmonary vascu-lature from both the arterial and venous reservoirsthroughout the equal-pressure-flow intervals. Lossesof fluid from both reservoirs were significantlygreater at 30 cm H2O than at 20 cm H2O (p < 0.05;see Fig. 4). A total of approximately 10 mL was lostfrom the reservoirs in the 30 cm H2O studies and 4mL in the 20 cm H2O studies. Although losses fromthe arterial reservoirs averaged more than lossesfrom the venous reservoirs, these differences werenot significant.

Uptake of FITC-Dextran and Ultrafiltration

During the loading interval, fractional concentra-tions of FITC-dextran (CFITC) in the venous outflowaveraged 7.0 ± 1.3% below those in the inflow in the20 cm H2O studies and 8.1 ± 1.3% below those inthe inflow in the 30 cm H2O studies (see upper pan-els of Figs. 5 and 6). Following the equal-pressure

period, they increased above inflow concentrationsin both the 20 and 30 cm H2O experiments (seemiddle panels of Figs. 5 and 6). Increases in FITC-dextran concentrations were attributed to ultrafiltra-tion [Eq. (1)]. Ultrafiltration averaged 0.43 ± 0.09mL during the equal-pressure intervals in the 20 cmH2O experiments and 0.56 ± 0.11 mL in the 30 cmH2O studies (this difference was not significant). Asindicated below, the unexpected failure to detectmore ultrafiltration at higher pressures may havebeen related to greater losses of perfusate from theexchange vessels of the lungs during the equal pres-sure interval.

Recovery of Arterial and Venous Indicators in theCollection Tubes

Perfusate from both the arterial and venous reser-voirs was found in the fluid flushed from the lungsafter the equal-pressure interval ended (Figs. 5 and6). Because the perfusate solutions in these reser-voirs were labeled with 14C-dextran and 3H-dextran,it was possible to determine how much of perfusateflushed from the vasculature at the end of the ex-periments were derived from each of these reservoirs(see bottom panels of Figs. 5 and 6). As expected,

Figure 4. Losses of perfusate from the arterial and ve-nous reservoirs during the equal-pressure interval. Losseswere greater at 30 cm H2O than at 20 cm H2O. Meanlosses from the arterial reservoirs were greater than thosefrom the venous reservoirs, but these differences were notsignificant.

Figure 3. Outflow rates from the pulmonary vein duringthe loading and flush intervals. Following the end of theequal-pressure intervals, there is a brief surge of fluidpresumably related to the prior distention of the vascularcompartment.


327

the venous indicator emerged from the lungs beforethe arterial indicator during the flush interval. Simi-lar quantities of the arterial and venous indicatorswere recovered in the 20 cm H2O experiments. Re-covery of the arterial indicator appeared to exceedthat of the venous indicator in the 30 cm H2O ex-periments, but this difference was not significant.Indocyanine green–labeled albumin was used to de-termine the arrival of the flush solution into the col-lection samples (Figs. 5 and 6) and to define thevolume of fluid that was flushed from the vascula-ture (the vascular volume).

As indicated in Fig. 7, more of the fluid that waspresent at the beginning of the equal-pressure inter-val (designated the “load” volume) was recovered inthe outflow during the flush interval in the 20 cmH2O studies than in the 30 cm H2O studies. Seventypercent of the perfusate flushed from the lungs in the20 cm H2O experiments was present in the lungs atthe start of the equal-pressure interval. The volumeof perfusate that entered the lungs from the arterial

and venous reservoirs during the equal-pressure in-terval (0.45 ± 0.10 mL) was not significantly differ-ent from the volume of transudation that had oc-curred (0.43 ± 0.11). In contrast, only 27% of theperfusate flushed at 30 cm H2O was in the lungsfrom the start of the equal-pressure interval. Move-ment of fluid out of the vasculature during theequal-pressure interval inevitably decreases calcu-lated transudation at these pressures (see Discus-sion).

The vascular volumes (calculated from the unex-trapolated indocyanine green data) were not signifi-cantly different at low and high pressures (1.43 ±0.05 mL at 20 cm H2O and 1.10 ± 0.23 mL at 30 cmH2O). It is quite possible that some of the perfusatewas not washed out of the pulmonary vasculature atthe end of the 30 cm H2O equal-pressure interval.This is suggested by the observation that the outflowpatterns of the flush indicator (indocyanine green–labeled albumin) were much less uniform at 30 cmH2O than at 20 cm H2O (Fig. 8). Extrapolation ofthe indocyanine green curves, by including Vresidualin Eqs. (4) and (5), modestly increased the calcu-lated vascular volumes to 1.67 ± 0.12 mL at 20 cmH2O and 1.28 ± 0.16 mL at 30 cm H2O.

<

Figure 5. Transudation and vascular filling at 20 cmH2O pressure. Upper panel: Fractional concentrations ofFITC-dextran in the venous outflow (CFITC) remainedslightly below those entering the lungs during the loadingperiod. Indicator concentrations during the flush period(designated as “After Stop-Interval”) are shown in themiddle and lower panels. Middle panel: Increases in CFITCover baseline levels in samples collected during the flushinterval indicated that transudation of fluid from the vas-culature had occurred during the equal-pressure interval.Bottom panel: Fractional indicator concentrations of theoutflow samples correspond to fluid derived from reser-voir I (containing the “load” perfusate), reservoir II (con-taining the flush perfusate, which is labeled with ICG),and the arterial and venous reservoirs (connected to thelungs during the equal-pressure interval, and labeled with3H- or 14C-dextran). All reservoirs contained the sameconcentrations of FITC-dextran. The initial samples con-tained perfusate that had been in the collection catheterduring the equal-pressure interval, accounting for highvalues of the load perfusate. Perfusate derived from thevenous reservoir was collected before that from the arte-rial reservoir. The quantities derived from these twosources were approximately the same. The appearance ofICG indicates the emergence of the flush solution into thecollection samples. Venous concentrations of ICG did notreach those entering the lungs from reservoir II, and ex-trapolation was used to calculate the full vascular volume,as indicated in the text.


328

Losses of Fluid into the Airspaces

At the end of the experiments, the airspaces werelavaged with a solution containing rhodamine-labeled dextran to determine if any fluid had enteredthe airspaces during the course of the experiments.Rhodamine-dextran concentrations in the fluid re-moved from the airspaces did not fall in the 20 cmH2O experiments, indicating that little or no fluidhad entered the airspaces during these studies (rho-damine concentrations in the fluid collected from theairspaces averaged 103 ± 1% those instilled into thelungs). However, concentrations of rhodamine in theairway fluid collected from the airspaces in the 30cm H2O experiments averaged only 71 ± 7% (p <0.05) those in the fluid instilled into the airways.This represented the recovery of 2.35 ± 0.76 mL of

edema fluid from the airspaces. FITC concentrationsin the lavage fluid were low in the 20 cm H2O studies(6.4 ± 0.3% of that in the perfusate). More of theFITC-dextran was recovered in the lavage fluid inthe 30 cm H2O experiments (17 ± 4% of that in theperfusate) (p < 0.05). This indicates leakage of per-fusate into the airspaces at these high pressures. Ap-proximately 40% of the FITC-dextran was removed(sieved) from the fluid entering the airspace com-partment during the 30 cm H2O equal-pressure in-tervals (calculated from the ratio of the fractionalconcentration of FITC-dextran in the airspace fluidto the fractional dilution of the rhodamine-dextranin the airspace).

None of the indicator in the flush solution, ICG-albumin, entered the airspaces in the 20 cm H2Olungs during the flush interval, but small amountswere found in the 30 cm H2O experiments (repre-senting 3.7 ± 1.7% of concentrations in reservoir II).

Losses of Fluid into the Jacketed Chamber

A total of 0.85 ± 0.12 mL of fluid was found at thebottom of the heated chamber at the end of the 20cm H2O experiments. Significantly more (3.79 ±0.99 mL) was found in the 30 cm H2O studies. Con-

Figure 7. Calculated filtration and the volumes of fluid inthe outflow samples that were derived from the four res-ervoirs shown in Fig. 1 (“Flush” 4 reservoir II, and“Load” 4 reservoir I). The flush volume defines the vas-cular volume at the end of the equal-pressure interval.Significantly more of the load solution was recovered inthe outflow samples in the 20 cm H2O experiments than inthe 30 cm H2O experiments (this is indicated by an as-terisk). The sum of the volumes of venous and arterialperfusate entering the lungs equaled the volume of fluidlost by filtration at 20 cm H2O. Losses of the load solutionat 30 cm H2O appeared to be associated with excess inflowof arterial perfusate into the vasculature during the equal-pressure interval.

Figure 6. Transudation and vascular filling at 30 cmH2O pressure (see Fig. 5 legend for details). Evidence forloss of the FITC-dextran between the inflow and outflowof the lungs is present during the load interval (upperpanel). Increases in FITC-dextran after the equal-pressureinterval were similar to those seen at 20 cm H2O, reflect-ing the fact that it was not possible to document moretransudation from the perfusate than in the 20 cm H2Oexperiments.


329

centrations of FITC-dextran were similar to those inthe fluid emerging from the outflow during the load-ing period (106 ± 2% in the 20 cm H2O experimentsand 96 ± 2% in the 30 cm H2O experiments).

When the radioactivities of the venous outflow, thefluid entering the external chamber, and that recov-erable from the airspaces were added together [Eq.(10)], the total recovery of 14C and 3H averaged 76± 9% and 69 ± 9%, respectively, in the 20 cm H2Ostudies and 57 ± 5% and 46 ± 8% in the 30 cm H2Ostudies. These differences were not significant.

DISCUSSION

The current study provides new information regard-ing the behavior of the vasculature of isolated, per-fused rat lungs at low and high distention pressures.

At 20 cm H2O, perfusate continues entering thelungs at 0.4 mL/min from the arterial and venousreservoirs but does not displace most of the perfusatethat was already in the pulmonary vasculature. Thisobservation is consistent with leakage from anasto-motic (bronchial) vessels that are at the arterial andvenous ends of the circulation. These anastomoticvessels remain disrupted when the lungs are excisedfrom the chest and the pulmonary artery and veinare cannulated. Although there is some controversyconcerning the relative number of anastomoses be-tween pulmonary arteries and veins and the bron-chial circulation, there appear to be significant anas-tomoses at both sites in the rat (3,12,13,17,19).Even if the lungs were not excised, leakage throughthese vessels into systemic veins would be expectedwhen arterial and venous pressures are equalized.

Retention of most of the perfusate in the pulmonaryvasculature despite persistent flow in the 20 cm H2Oexperiments suggests that pressures at the arterialand venous sites of leakage are nearly equal. Therecovery of similar volumes of perfusate from thearterial and venous reservoirs in the flush period(Fig. 5, bottom) is consistent with this hypothesis.This hypothesis can be conceptualized with the sche-matic diagram shown in Fig. 9(A). At low pressures,there is some loss of fluid by filtration from the ex-change vessels, which can be assessed in these ex-periments by an increase in the concentration ofFITC-dextran (note increased shading in the figure).This loss of volume is replenished by a small amountof fluid entering from the arterial and venous reser-voirs, but most of the perfusate that was already inthe lungs would remain there. The total amount ofarterial and venous perfusate that was recoveredfrom the pulmonary vasculature during the equal-pressure interval was about the same as the amountof filtration that had occurred in the 20 cm H2Oexperiments.

Significantly greater flows (12 mL/min) persistedduring the equal-pressure periods at 30 cm H2Othan at 20 cm H2O (Fig. 4). Furthermore, signifi-cantly less of the original perfusate remained in theexchange vessels at the end of the 30 cm H2O ex-periments. Replacement of the original perfusate inthe pulmonary vasculature with perfusate from thearterial and venous reservoirs could reflect loss of theoriginal perfusate into the anastomotic vessels. Forexample, greater losses of perfusate from venousanastomoses will result in flow of perfusate from thearterial to venous vessels and movement of the origi-nal perfusate out of the pulmonary vasculature [Fig.9(B)]. Note that this would result in losses of the

Figure 8. Washout of the pulmonary vasculature by theflush solution (containing indocyanine green). ICG desig-nates the concentration of ICG and the value of 1 − CICGindicates the fraction of the collected perfusate that was inthe vasculature and dead space at the end of the equal-pressure interval, just before the flush was commenced. Asemilogarithmic plot was used to permit extrapolation ofthe outflow curves beyond the last sample (see Materialsand Methods). Washout was more uniform at 20 cm H2Othan at 30 cm H2O.


330

concentrated FITC-dextran and ultrafiltration fromthe exchange vessels would be underestimated. Sig-nificantly more fluid leaked into the lung chamber at30 cm H2O than at 20 cm H2O. FITC-dextran con-centrations in this fluid were no different from thosein the perfusate, indicating leakage from the vascu-lature rather than ultrafiltration.

Additional fluid was lost into the airspaces in the 30cm H2O but not in the 20 cm H2O experiments.FITC-dextran concentrations in the airspace fluid inthe 30 cm H2O experiments were approximately60% of those in the perfusate. Sieving of FITC-dextran could reflect the properties of the pulmonaryendothelium or epithelium to allow some movementof this macromolecule. Because there was no move-ment of fluid or FITC-dextran into the airspaces at20 cm H2O, it is also possible that the FITC-dextranfound in the airspaces was associated with leakagethrough disrupted membranes that normally sepa-

rate the vascular and airspace compartments. As il-lustrated in Fig. 9(C), leakage from the pulmonaryvessels would also result in underestimation of ultra-filtration.

Losses of the original perfusate from both anasto-motic and pulmonary exchange vessels could be re-sponsible for the failure of calculated filtration at 30cm H2O to significantly exceed that at 20 cm H2O.Ultrafiltration may have also occurred in bronchialvessels, but this could not be detected in these ex-periments because perfusate in the bronchial vesselswas presumably not flushed into the pulmonaryveins at the end of the experiments.

Movement of fluid from the arterial to the venousend of the vasculature during the equal-pressure in-terval could, in theory, contribute to our earlier ob-servation that transudation appeared to occur atmore venous sites than those at which exchange of

Figure 9. Schematic diagram of some of the possible effects of persistent flow during “equal-pressure-flow” measure-ments of filtration in isolated lungs. Panel A, 20 cm H2O experiments: Elevations of intravascular pressure promoteultrafiltration of fluid from the exchange sites in the pulmonary vasculature. This causes FITC-dextran concentrationsto increase (indicated by increased shading; FITC-dextran was included in all of the perfusion solutions). Freshperfusate enters the pulmonary vasculature from both the arterial and venous reservoirs and replaces the volume lostby ultrafiltration. Most of the perfusate entering the lungs from the arterial and venous reservoirs is lost throughprecapillary and postcapillary anastomotic (bronchial) vessels. Flow remains minimal in the exchange vessels. PanelsB and C: Possible losses of perfusate from the pulmonary vasculature at 30 cm H2O. Panel B: If there is more loss ofperfusate from the venous than the arterial anastomoses and local venous pressures fall below those in arterial vessels,perfusate in the exchange site will tend to flow toward more venous portions of the circulation and may be lost throughvenous anastomoses. Panel C: Some of the concentrated perfusate is lost through defects in the pulmonary exchangevessels.


331

3HOH from the airspaces occurred (9). In the earlierexperiments, we instilled the 3HOH at the end of theequal-pressure-flow period and found that increasesin concentrations of FITC-dextran appeared in ve-nous outflow samples before 3HOH was detected.However, our previous study was conducted at lowerpressures, and we could find no evidence for prefer-ential movement of the arterial indicator into thevasculature at these pressures in the current study.Nevertheless, it may be preferable to also instill3HOH into the airspaces at the beginning of theequal-pressure-flow interval, to compensate for anyflows that might be taking place during the equal-pressure-flow interval.

Concentrations of the arterial and venous indicators(14C and 3H-dextran) were similar in the collectionsamples at 20 cm H2O (Fig. 5). At 30 cm H2O,average concentrations of the arterial indicator ap-peared to be greater than those of the venous indi-cator (Fig. 6) and there appeared to be greater lossesfrom the arterial reservoir (Fig. 4). These differenceswere not quite significant, but it would appear thatvenous filling of the vasculature during the equal-pressure interval was not greater than arterial filling.This observation was unexpected because venouspressures were lower than those in the artery prior tothe equal-pressure-flow interval and there was con-sequently a greater increase in venous pressure thanin arterial pressure when the equal-pressure-flowperiod began. Greater filling at the venous end mighttherefore have been anticipated. It is possible thatmore perfusate was lost from venous sites [Fig.9(B)], presumably into the chamber and airspaces.If this occurred, fluid from the arterial reservoircould replace much of that in the vasculature.

It is also possible that distention of the microcircu-lation increased leakage of dextran through themembranes that separate the pulmonary vasculatureand the extravascular compartments. In otherwords, the reflection coefficient of the membranes todextran may have been reduced. This would alsotend to reduce increases in vascular FITC-dextranconcentrations associated with transudation.

Some of the loss of the original “load” perfusate thathad entered the lung prior to the equal-pressure in-terval at 30 cm H2O may have been more apparentthan real. Flushing of the lungs seemed to be lessefficient at 30 cm H2O than at 20 cm H2O. As in-dicated in Fig. 8, the washout curves of indocyaninegreen were more irregular. Consequently, some ofthe “load” perfusate that remained in the vascula-ture may have remained in the vasculature at the

end of the experiments. Incomplete flushing mayhave also been responsible in part for the fact thatthe vascular volume calculated from the ICG-albumin curve was not greater at 30 cm H2O than at20 cm H2O. It is possible that interstitial edema,cellular swelling, or alveolar filling associated withinjury in the equal-pressure-flow conditions at 30cm H2O may have reduced the vascular volume at30 cm H2O (8,18). Irregular flushing of indicatorsfrom different regions of the lungs will tend to makeit more difficult to localize the sites of exchange andtransudation along the length of the pulmonary vas-culature.

Evidence was obtained that the volume of at leastsome portion of the pulmonary vasculature in-creased during the equal-pressure intervals: flow in-creased at the venous outflow when venous pressureswere decreased at the end of the equal-pressure in-terval (Fig. 3). It was not possible to distinguish be-tween vascular recruitment and distention. Increasesin the total volume of the pulmonary vasculatureshould not affect the calculated volume of ultrafil-tration, provided that none of the perfusate in theexchange vessels is lost.

Several innovations incorporated in this study de-serve mention. The most important change in thisequal-pressure-flow study was the incorporation ofdifferent dextran labels in the arterial and venousreservoirs. This made it possible to detect how muchof the perfusate that was originally in the lungs re-mained there at the end of the equal-pressure-flowinterval and the relative sites of the anastomotic flowand filtration. The following innovations were intro-duced which may prove useful in future studies ofthe pulmonary circulation in isolated lungs.

1. Cannulation of the pulmonary vein (there is acommon vein that enters the left atrium in rats)eliminated artifacts associated with left atrial dis-tention when outflow pressures were increased.

2. The overflow procedure allowed us to maintainconstant and very similar arterial and venouspressures throughout the experiments.

3. Measurement of pressures in small cathetersplaced in the storage cylinders provided continu-ous measurements of flow into the lungs from thearterial and venous reservoirs during the equal-pressure-flow interval.

4. Indocyanine green in the flush solution was usedto estimate the vascular volume at the end of theexperiments.

5. Movement of the collection rack was regulatedwith a drop counter. This resulted in the collec-


332

tion of virtually all of the fluid emerging from thevenous outflow.

6. Each of the collection samples were weighed andthe time duration of each collection sample wasrecorded. The outflow rate that corresponded toeach of the collection samples was calculatedfrom these data.

7. Intratracheal lavage with rhodamine-dextranpermitted detection of leakage of fluid into theairspaces.

8. The thermostated chamber in which the lung wasplaced assured constant temperatures during theequal-pressure-flow interval.

Outflow concentrations of FITC-dextran wereconsistently slightly below those in the arterial in-flow in these experiments (top panels, Figs. 4 and 5).It is possible that this reflects chemical alteration(reduction?) of the FITC-moiety or adhesion to pro-tein lining the vasculature. Similar changes in theconcentrations of other indicators such as Evansblue and methylene blue have been observed and ithas been suggested that reduction or protein bindingby the endothelium is responsible for this phenom-enon (1,5,6,16,20). The actual causes for smalllosses of Evans blue from the vasculature in some ofthese studies remain uncertain. In our previousstudy, we found that increases in FITC-dextran rela-tive to baseline venous concentrations (rather thanarterial concentrations) were similar to increases inconcentrations of albumin in the perfusate, suggest-ing that increases in FITC-dextran concentrationsabove venous baseline levels can be reliably used toestimate transudation (9). Furthermore, the similar-ity between the calculated volume of fluid lost byultrafiltration and the volume of perfusate that en-tered the vasculature from the arterial and venousreservoirs during the equal-pressure interval suggestthat this method of determining transudation is ap-propriate. Regardless of the precision of the ultrafil-tration measurement in these experiments, it wouldappear that the location of anastomotic losses in the20 cm H2O studies occurred through vessels thatwere proximal or distal to the exchange sites in thelungs.

These studies indicate that despite persistent flow ofperfusate into the lungs from the arterial and venousreservoirs, there is very little flow in the pulmonaryexchange vessels when pressures are kept at 20 cmH2O. This is presumably related to the fact thatanastomotic flows occur at sites that are not in theexchange regions of the lungs. The absence of sig-nificant flow in the exchange vessels makes it pos-sible to measure exchange and filtration in equal-

pressure studies in which lungs are exposed to arte-rial and venous pressures of 20 cm H2O. At 30 cmH2O, there are significantly greater losses of perfus-ate into the airspaces and chamber, vascular wash-out is less predictable, and it is difficult to measureultrafiltration.

ACKNOWLEDGMENT

We would like to thank Dr. Christopher Dawson forhis helpful advice in these studies.

REFERENCES

1. Bongard RD, Krenz GS, Linehan JH, Roerig DL,Merker MP, Widell JL, Dawson CA. (1994). Reduc-tion and accumulation of methylene blue by the lung.J Appl Physiol 77:1480–1491.

2. Chang RSY, Silverman P, Effros RM. (1979). Vascu-lar and extravascular compartments of the isolatedperfused rabbit lung. J Appl Physiol Respirat EnvironExercise Physiol 46:74–78.

3. Charan NB, Turk GM, Dhand R. (1984). Gross andsubgross anatomy of bronchial circulation in sheep. JAppl Physiol Respirat Environ Exercise Physiol 57:658–664.

4. Drake R, Gaar KA, Taylor AE. (1978). Estimation ofthe filtration coefficient of pulmonary exchange ves-sels. Am J Physiol 234 (Heart Circ Physiol 3):H266–H274.

5. Dean PD, Watson DH. (1979). Protein purificationusing immobilized triazine dyes. J Chromatogr 165:301–316.

6. Effros RM, Schapira R, Presberg K, Ozker K, JacobsER. (1998). Equal pressure-flow studies of solute up-take in rat lungs J Appl Physiol 85:986–992.

7. Harris NR, Parker RE, Pou NA, Roselli RJ. (1992).Canine pulmonary filtration coefficient calculatedfrom optical, radioisotope, and weight measurements.J Appl Physiol 73:2648–2661.

8. Kraft SA, Fujishima S, McGuire GP, Thompson JS,Raffin TA, Pearl RG. (1995). Effect of blood andalbumin on pulmonary hypertension and edema inperfused rabbit lungs. J Appl Physiol 78:499–504.

9. Lin W, Jacobs E, Schapira RM, Presberg K, EffrosRM. (1998). Equal pressure-flow studies of distribu-tion of filtration in rat lungs. J Appl Physiol 84:47–52.

10. Lunde PKM, Waaler BA. (1969). Transvascular fluidbalance in the lung. J Physiol 205:1–18.

11. Malvin RL, Wilde WS. (1973). Equal pressure-flowtechnique. In: Handbook of Physiology, Section 8: Re-nal Physiology (J Orloff and RW Berliner, Eds.)American Physiological Society. Washington, DC.119–143.

12. McLaughlin RF Jr. (1983). Bronchial artery distribu-tion in various mammals and in human. Am RevRespir Dis 128:557–558.

13. Miller WS. (1947). The Lung. Charles C. Thomas.Springfield, IL.


333

14. Oppenheimer L, Unruh HW, Skoog C, Goldberg HS.(1983). Transvascular fluid flux measured from in-travascular water concentration changes. J ApplPhysiol Respirat Environ Exercise Physiol 54:64–72.

15. Pilati CF, Maron MB. (1986). A technique to measurethe reflection coefficient using endogenous vascularindicators. Microvasc Res 32:255–260.

16. Pfister RE, Ramsden CA, Neil HL, Kyriakides MA,Berger PJ. (1999). Errors in estimating lung liquidvolume in fetal lambs when using radiolabeled serumalbumin and blue dextran. J Appl Physiol 87:2366–2374.

17. Schraufnagel DE. (1989). Microvascular casting of

the lung: bronchial versus pulmonary artery filling.Scanning Microscopy 3:575–578.

18. Teplitz C. (1979). Pulmonary cellular and interstitialedema. In: Pulmonary Edema (AP Fishman and EMRenkn, Eds.) American Physiological Society. Wash-ington, DC. 97–112.

19. Wagner EM, Mitzner W, Brown RH. (1999). Site offunctional bronchopulmonary anastomses in sheep.Anat Rec 254:360–366.

20. White P Jr, Brower R, Sylvester JT, Permutt T, Per-mutt S. (1993). Factors influencing measurement ofprotein reflection coefficient by filtered volume tech-nique. J Appl Physiol 74:1374–1380.


334

relationship of ultrafiltration and anastomotic flow in isolated rat lungs

Documents