Pathogenesis of Malignant Ascites Formation: Initiating ... · cavity, increased by 2- to 40-fold in ascites tumor-bearing animals as compared with peritoneal fluid volume- and protein-matched

[CANCER RESEARCH 53. 2631-2643. June I. 1993]

Pathogenesis of Malignant Ascites Formation: Initiating Events That Lead to FluidAccumulation1

Janice A. Nagy,2 Kemp T. Herzberg, Jane M. Dvorak, and Harold F. Dvorak

Departments nf Pathology, Beth Israel Hospital and Han-ard Medical Schon/. Boston. Massachusetts 02215

ABSTRACT

Initiating events leading to the accumulation of malignant ascites in theperitoneal cavity were investigated in two syngeneic transplantable murine ascites-producing tumors, MOT mouse ovarian tumor and the TA3/Stmammary carcinoma. The transport of two tracers. I25l-labeled humanserum albumin (125I-HSA) and 5lCr-labeled red blood cells (5lCr-RBC),

into and out of the peritoneal cavity was studied at early times after i.p.tumor cell injection, prior to abundant fluid accumulation, and at intervals of 5 to 360 min after i.v or i.p. tracer injection. Tracer influx andefflux rates were estimated from the mass of tracer passing into or out ofthe peritoneal cavity following a bolus injection of tracer into either theblood or the peritoneal cavity. Efflux of I25I-HSA from the peritoneal

cavity was markedly reduced (3- to 5-fold) within 1 day of i.p. injection of

either type of tumor cell. Significantly reduced efflux preceded any increase in tumor cell number and by itself did not induce peritoneal fluidaccumulation. I2SI-HSA tracer influx from plasma to peritoneal fluid did

not increase detectably until 5 to 7 days after tumor cell injection, whenthe tumor cell number had increased by 10- to 100-fold. Only at relatively

late stages of ascites tumor growth, when the flow rate into the peritonealcavity had increased relative to the flow rate out of the peritoneum, wasthere net peritoneal fluid accumulation. Thus, increased influx, in additionto impaired efflux, were required for malignant ascites accumulation.Following i.p. injection, the efflux rates of ' '5I IIS Valways exceeded those

of MCr-RBC, even in ascites tumor-bearing animals. Furthermore,1-"1-1IS \ tracer disappeared from the peritoneal cavity more rapidly thanit appeared in the plasma, suggesting that ' "I-IIS \ moves more rapidly

through the channels by which 5lCr-RBC egress from the peritoneum

(primarily diaphragmatic lymphatics) and/or has access to additionalpathways not open to '( i-KIH . Finally, flow rates into and out of the

blood and peritoneum were used to obtain kinetic parameters that characterized tracer transport: k,, the rate constant for tracer transport fromthe blood to the peritoneum; k2, the rate constant for tracer transportfrom the peritoneal cavity to the blood; and A,,,the rate constant for tracertransport from the peritoneal cavity to surrounding interstitial tissue.Using these rate constants together with the differential equations thatgovern a three-compartment model (plasma or blood, peritoneal cavity,

and the extravascular space of all other organs), theoretical influx andefflux curves were generated and compared with the experimentally determined values for tracer concentration in plasma (or blood) and peritoneal fluid at varying intervals after tracer injection. This simple three-

compartmental analysis successfully modeled influx as well as efflux datafor both tracers.

INTRODUCTION

Abnormal accumulation of a plasma protein-rich exúdate commonly accompanies tumor growth in serous cavities (1-8). However,

the mechanism(s) responsible for malignant ascites accumulation areincompletely understood. Studies of solid tumors have shown that thelocal microvasculature is leaky, and it has been suggested that tumorascites fluid accumulates as the result of hyperpermeability of theblood vessels that line serous cavities (1-4, 7. 8). In support of this

Received 12/18/91; accepted 3/29/93.The costs of publication of this article were defrayed in part by the payment of page

charges. This article must therefore be hereby marked advertisement in accordance with18 U.S.C. Section 1734 solely to indicate this fact.

1Supported by NIH Research Grants CA2847I and CA50453, under terms of a

contract from the National Foundation for Cancer Research and by the B1H PathologyFoundation. Inc.

2 To whom requests for reprints should be addressed, at Department of Pathology. Beth

Israel Hospital. Research East. 330 Brookline Avenue. Boston, MA 02115.

view, many tumors, both solid and ascites, secrete a protein mediator.VPF1 (9. 10). that greatly enhances the permeability of normal blood

vessels to circulating macromolecules (8, 11).Alternatively, it has been argued that the ascites accumulation of

malignancy may result from the impedance of peritoneal fluid drainage (12-17). Soluble proteins and particles such as colloidal carbon

and erythrocytes rapidly egress from the peritoneal cavities of normalanimals, predominantly by way of lymphatic channels concentrated inthe diaphragm (12, 18-23). In contrast, both clinical and experimental

evidence indicate that diaphragmatic lymphatic drainage from theperitoneal cavity is reduced in malignant ascites (12-17).

Previously we investigated the influx and efflux of macromolecularFITC-D into and out of the peritoneal cavities of mice bearing either

of two transplantable ascites tumors: the MOT mouse ovarian tumorand the TA.VSt breast carcinoma (24). These studies were performed7-10 days after i.p. tumor cell injection, i.e., after substantial ascites

fluid had already accumulated. Influx, as measured by k,, the rateconstant for the transfer of FITC-D from the plasma to the peritonealcavity, increased by 2- to 40-fold in ascites tumor-bearing animals ascompared with peritoneal fluid volume- and protein-matched control

animals. Efflux, as measured by k2. the rate constant for the transferof FITC-D from the peritoneum to plasma, was reduced 5- to 50-foldin ascites tumor-bearing animals. Thus, at relatively late stages of

tumor growth, when abundant ascites was already present, accumulation of macromolecules and fluid in the peritoneal cavity reflectedboth increased influx and impaired efflux.

These last experiments provided new information regarding themaintenance of tumor ascites but did not define the initiating eventsleading to ascites generation. Once protein-rich fluid is present within

a serous cavity, the increased intracavity hydrostatic and oncotic pressures inevitably affect influx and efflux rates (25-27). Also at these

late times, we now know, multiple secondary changes have takenplace in the anatomy and physiology of the peritoneal wall and diaphragm, including breaching of the mesothelial barrier, tumor cellattachment, fibrin deposition, angiogenesis, and fibrosis (28, 29).These alterations must also have profoundly affected the balance ofmacromolecular transport into and out of the peritoneal cavity.

To elucidate the pathogenesis of malignant ascites formation, therefore, it was necessary to investigate tracer influx and efflux at theearliest stages of tumor growth, prior to net fluid accumulation andbefore the development of secondary alterations in the peritoneallining. For this study, we elected to use the same pair of syngeneic,transplantable murine tumors used earlier (24). However, because wehad found that influx and efflux rate constants were only modestlyinfluenced by tracer size over a wide range (Mr 70.0()()-2.()(X).(X)0).

there was no need to use a multisized family of tracers, such as theFITC-dextrans we had used previously (24). Therefore, we used I2SI-labeled human serum albumin (I2SI-HSA; M, 69,000) as our primary

tracer because of its physiological relevance as the major protein ofplasma and of malignant ascites (30, 31 ). Because soluble macromolecular tracers, such as albumin, have been reported to exit the peritoneal cavity by traversing the peritoneum (22, 32, 33) in addition toexiting by the diaphragmatic lymphatics (22, 34—37),we also investigated a paniculate tracer. ""'Cr-RBC, which is thought to exit the

1The abbreviations used are: VPF. vascular permeability factor; FITC-D. fluorescein-

labeled dextrans; HSA. human serum albumin; BSA, bovine serum albumin.

2631

Research. on October 17, 2020. © 1993 American Association for Cancercancerres.aacrjournals.org Downloaded from

PATHOCENESIS OF MALIGNANT ASCITES

peritoneal cavity solely by way of draining lymphatics and predominantly those residing in the diaphragm (18, 21, 23, 36, 38^0).

We now report that efflux from the peritoneal cavity of 125I-HSAand 5lCr-RBC is markedly reduced within 1 day of i.p. tumor cell

injection. Significant reduction of efflux preceded a detectable increase in tumor cell number, was not attributable to blockage ofperitoneal lymphatics by tumor cells, and by itself did not provokeperitoneal fluid accumulation. On the other hand, '-5I-HSA tracer

influx did not increase detectably until 5-7 days after tumor cellinjection, when tumor cell number had increased 10- to 100-fold: only

at this relatively late stage, when tracer influx into the peritonealcavity had increased substantially, did peritoneal fluid begin to accumulate.

MATERIALS AND METHODS

Animals, Tumors, Tracers, and Other Reagents. The MOT and TA3/Sttransplantable ascites-producing tumors were passaged weekly in the perito

neal cavities of syngeneic C3Heb/FeJ and A/Jax mice, respectively (41 ). Human serum albumin (Sigma Chemical Co.. St. Louis. MO) was iodinated withlodo-Gen (42) (Pierce Chemical Co., Rockford, IL) to a specific activity of0.02-0.05 mol '"I/mol HSA; >97ft of '-'I-HSA was precipitated by 10%

trichloroacetic acid. Freshly collected mouse RBC were labeled with "Cr

(sodium chromate; Dupont-NEN, Boston, MA) (43). Total protein in cell-freeascites was measured by the dye-binding method of Bradford (44) with BSA

as the standard. The albumin content of ascites fluid was determined by thebromocresol green dye-binding assay (Sigma Chemical Co.) with mouse al

bumin as standard.r\|icriiiiciit:il Design. Syngeneic A/J or C3Heb/FeJ mice (4-6 weeks old.

15-20 g body weight) were studied at various intervals after i.p. injection of1.0 X IO6 TA.VSt or MOT tumor cells. Influx and efflux measurements wereperformed by injecting either 2.5 x IO6 cpm of I25I-HSA (in 0.2 ml of saline)or 1.0 X 10" cpm of 5lCr-RBC (in 0.5 ml of saline) i.v. for influx studies and

i.p. for efflux studies. At various time intervals thereafter, ascites tumor-

bearing or control animals were anesthetized with ether, and blood sampleswere collected by retroorbital puncture into a known volume of heparin forpreparation of platelet-poor plasma (centrifugation at 15,600 x g. 10 min.4°C).Animals were then sacrificed with ether and exsanguinated. Two ml ofHanks' balanced salt solution were injected i.p., and the contents of the peri

toneal cavity were mixed by kneading. The peritoneal cavity was then openedby a small ventral incision and its contents recovered to the fullest extentpossible by syringe. The total peritoneal fluid volume was recorded, and tumorcells were counted. An aliquot of peritoneal fluid was removed for radioactivecounting. Peritoneal fluid was then centrifuged (160 x g, 10 min, roomtemperature), and the volume of cell-free supernatant was determined. Aliquotsof blood, plasma, peritoneal fluid, and cell-tree peritoneal fluid were analyzedfor '"I-HSA or "Cr-RBC content by gamma counting. In addition, whens'Cr-RBC was used as a tracer, the spleen was excised, and the radioactivity

present was determined as an indication of the extent of clearance of damagedRBC.

Control animals were of two types: (a) age- and sex-matched normal miceof either strain and (b) similar animals (C-BSA) that had been injected i.p. with2 ml (TA3/St) or 5 ml (MOT) of 5<7rBSA in saline immediately prior to tracer

injection to match the volume and protein content of ascites fluid in tumor-

hearing animals at 7 days after i.p. injection of tumor cells (24).Measurement of Tracer Flow Rates. Transcompartmental tracer flow

rates were calculated from the mass of tracer passing between the peritoneal

tracer iv tracer Ip

1excretion 4—¿�plasma

orbloodk1X *2peritoneal cavity

1 L

Fig. 1. Three-compartment model consisting of plasma (or blixid). peritoneal cavity,and the extravascular space of all other tissues. Rate constants ki through kf, (min"1)

describe rates of tracer transport between compartments, as well as excretion. Tracer isintroduced into the system as a single bolus injection, either into the blood (i.v.l for influxexperiments or into the peritoneal cavity (i.p.) for efflux experiments.

the peritoneal cavity to account for the peritoneal radioactivity. We then plottedthis volume as a function of time after tracer injection and calculated the tracerflow rate. i.e.. the slope of the line (ul/min). by linear regression.4 In addition

to the rate of tracer appearance in the peritoneum we also determined the rateof tracer disappearance from the plasma (or blood) as the volume of plasma (orblood) that had moved out of the vascular space to account for the loss inplasma (or blood) radioactivity over the same time period.

Efflux rates were calculated from the mass of tracer passing out of theperitoneal cavity and into the plasma (or blood) following a bolus i.p. injection.From the quotient of the radioactivity (cpm) recovered in the plasma (or blood)at any particular time (5 to 180 min) after i.p. injection of tracer and the initialconcentration (cpm/ul) of tracer in the peritoneal fluid, we calculated thevolume (ul) of peritoneal fluid which had moved into the vascular compartment to account for the plasma (or blood) radioactivity. As above, we plottedvolume versus time and determined tracer flow rate by linear regression. Inaddition to the rate of tracer appearance in the blood we also determined therate of tracer disappearance from the peritoneal cavity as the volume ofperitoneal fluid that had moved out of the peritoneal cavity to account for theloss in peritoneal fluid radioactivity.

Mathematical Model and Determination of Rate Constants. To analyzethe data further we used a simplified mathematical model (24. 49) comprisingthree compartments: the intravascular space (plasma when I25I-HSA wastracer, whole blood when 5lCr-RBC was tracer): the peritoneal cavity (perito

neal fluid); and all extravascular spaces other than the peritoneal cavity (Fig.1). One major purpose of using tracer methods for kinetic studies in vivo is tocalculate intercompartmental flow rate constants which cannot be measureddirectly. Compartmental analysis is based on the assumption that specific poolscan be identified and that discharge of tracer from specific pools can bedescribed by exponential equations. The fraction of tracer lost from a givenpool per unit time is defined as the "fractional rate constant" or simply the rate

constant. Once the flow rate is known, then the rate constant is calculable asthe quotient of the flow rate and the volume of the pool from which the traceris being lost (49).

To determine the rate constant for influx. k¡.we calculated the quotient ofthe experimentally determined tracer flow rates (ul/min) into the peritonealcavity from the plasma (or blood) and the plasma (or blood) volume (49).Plasma and blood volumes were taken as the average of available publishedvalues for the mouse (50). i.e.. 51.8 ml plasma and 77.8 ml blood/kg body

cavity and the vascular space following either i.v or i.p. injection (15, 23, 45. we'Snt- Body weiSnt was corrected for the weight of the ascites fluid present

46). The primary assumption inherent in this method, i.e.. that tracer-contain

ing peritoneal fluid or plasma is removed from the peritoneal cavity or vascularspace without concentration or dilution, is well supported (26. 47). Furthermore, available evidence suggests that lymphatics have no ability to concentrate proteins (48).

Influx rates for I25I-HSA (or s'Cr-RBC) were calculated from the mass of

tracer passing into the peritoneal cavity following a bolus i.v. injection. Fromthe quotient of the radioactivity (cpm) recovered in the peritoneal cavity at anyparticular time (5 to 180 min) after i.v. injection of tracer and the initialconcentration (cpm/ul) of I25I-HSA in the plasma (or 5lCr-RBC in the blood),

we calculated the volume (ul) of plasma (or blood) which had extravasated into

at different intervals after tumor cell injection, taking the specific gravity ofascites fluid as 1.02 (51 ).

J It is possible thai our treaiment of the experimental data may underestimate the true

values for transcompanmental flow rates in that flow rates were based solely on linearregression analysis, and the elimination rale of tracer from the destination compartmentwas not included in these determinations. For example, in efflux sludies of the rate ofappearance of tracer in the bloodstream following i.p. injection, once ihe labeled tracergained entry inlo the bloodstream it would begin filtering back out into the peritonealcavity and other tissues; therefore, based on the tracer clearance curves, our values for rateof appearance in the vasculature would be underestimated by ~15*. Nevertheless, these

experimentally determined flow rates provide useful qualitative information on changes intracer transport that accompany ascites tumor growth.

2632

Research. on October 17, 2020. © 1993 American Association for Cancercancerres.aacrjournals.org Downloaded from

PATH<X;Õ;NF.SISoí-MALIÕÕNANTASCITKS

Analogously, (he rate constant for efflux, k;, was calculated as the quotientof the experimentally determined tracer flow rates out of the peritoneal cavityand into the plasma (or hlood) and the peritoneal fluid volumes (49). The verysmall volumes of peritoneal fluid present in control animals were measured bythe indicator dilution method (52). Briefly. |:M-HSA (5 x If)4 cpm/ml) in 1.0

ml saline was injected i.p. immediately after euthani/ation. The abdomen waskneaded to mix the peritoneal fluid, and the peritoneal cavity was opened andits contents recovered. The radioactivity present in an aliquot (O.I ml) ofrecovered fluid, (cpm/ml ),mai. was used in conjunction with the radioactivitypresent in the original stock solution (cpm/ml )mll,.,i.to determine the peritonealvolume according to liquation I:

(cpm/ml )riru|= (cpm/ml )ini (I)

where - is the peritoneal fluid volume in ml. At intervals prior to uscites

accumulation in tumor-hearing animals |/.<<.. for time points up to 3 days

(TA3/SO or 5 days (MOT)|. the peritoneal fluid volume was taken as the sumof the volume in which tracer was injected plus the volume of peritoneal fluiddetermined in control animals by the indicator dilution method. At later timesin ascites tumor-bearing mice, peritoneal fluid volume was taken as the amount

of fluid that could be recovered from the peritoneum: these values were verysimilar to those obtained in ascites tumor-bearing mice by the indicator dilution

method.The quotient of the experimentally determined rate of tracer disappearance

from the plasma (or hlood) and the plasma (or blood) volume was used todetermine the rate constant for tracer disappearance from the blood (equivalentto the sum of the rate constants k,. k\, and Ay, see Fig. 1). Values of the sumof kt and i, were then calculated by subtraction of the corresponding values ofk,. Finally, the quotient of the experimentally determined rate of tracer disappearance from the peritoneal cavity and the peritoneal fluid volume was usedto determine the rate constant for tracer disappearance from the peritonealcavity (equivalent to the sum of the rate constants k? and kh: see Fig. 1). Valuesof kf, were then calculated by subtraction of the corresponding values of ¿2.

Comparison of Experimental Data to Model-predicted Results. Using

the experimentally determined rate constants, the set of simultaneous differential equations for the rate of change of fraction of tracer in the three compartments was integrated numerically by using the Crank-Nicholson finite

difference scheme (53) to generate calculated data sets for influx and effluxthat characterized macromolecular transport between the plasma (or hlood) andthe peritoneal cavity (24). The calculated influx and efflux curves were thencompared to the experimental data. The experimental values used for comparison were obtained as follows. The amounts of either tracer present in plasma(or blood) and in peritoneal fluid were measured at successive intervals following i.v. or i.p. injection. For both influx and efflux experiments, and at eachtime interval, the percentage of the administered dose of tracer present in

plasma (or hlood) and peritoneal fluid was plotted as a semilogarithmic function against time after tracer injection. Data from the plasma (or blood) andperitoneal compartments were compared graphically with the influx and effluxcurves generated using the rate constants governing intercompartmental transfer of both tracers. For both tracers, the values of k,. k2. and kh and the sum ofk\ and Jt5 were calculated directly from the experimentally determined flowrates. For the |:5I-HSA tracer, as in our previous study with FITC-D (24). the

values of *4 and A>were constrained to be consistent with published values (23,35. 54-56) and were held constant for normal and ascites tumor-bearing mice.For 5'Cr-RBC. the values of kt. k¿.and A>were set equal to /ero (see below).

RKSULTS

Growth of Tumor Cells and Ascites Fluid Accumulation as aFunction of Time

Following the i.p. injection of 1 X IO6tumor cells, both TA3/St and

MOT tumors grew in logarithmic fashion for ~4 days and gradually

plateaued thereafter (Fig. 2, A and B). Net peritoneal fluid accumulation, however, did not begin for 4 days (TA3/SI) or 6-8 days (MOT),

but thereafter increased exponentially. Fluid accumulation plateauedin TA3/St tumor animals at 2-3 ml by day 6, and animals generally

died after day 8: in contrast, fluid continued to increase steadily inMOT-bearing mice for nearly a month and to a volume of —¿�25ml.

The total protein concentrations of tumor ascites fluids increased insigmoid fashion for both tumors from starting levels of —¿�10-15mg/ml to plateau values of >40 mg/ml by day 4 (TA3/SO and day 8

(MOT) (Fig. 2, C and D). Ascites levels of albumin paralleled totalprotein levels for both tumors and always remained at —¿�50%of thetotal protein concentration, plateauing at 20-25 mg/ml (data not

shown). Plasma protein levels decreased progressively but graduallyin ascites tumor-bearing animals, tailing to values of 80-85% of

control animals by day 7 (TA3/SD and day 21 (MOT) (Fig. 2, C and D).

Influx and Efflux Studies with I25I-HSA as Tracer

Flow Rates. Fig. 3 shows the influx and efflux of '~5I-HSA into

and out of the peritoneal cavity as a function of time following i.v. ori.p. injection of tracer on various days following tumor cell inoculation for TA3/St and MOT ascites tumors, respectively. Tracer flow isplotted as the volume (|jl) of plasma (or peritoneal fluid) transportedto the peritoneal cavity (or blood) following an i.v. (or i.p.) injectionof tracer. In all cases influx and efflux were linear over the timeperiods tested (up to 60 or 180 min), and the data were fit by linear

TA3/St MOT

Fig. 2. Growth characteristics of TA.VSt andMOT ascites tumors. A and H. numbers of tumorcells (D) and volumes ot ascites fluid (•)are plot-led against lime for TA.VSl and MOT ascites tumors, respectively, following i.p. inoculation of l xH)*' cells. C and D. protein content ot" peritoneal

fluid •¿�and plasma (O) for TA.VSt and MOTascites tumors, respectively. ß«rv,SD.

5

I

12 16 20 24 28

Days after ip tumor cell inoculation

2633

Research. on October 17, 2020. © 1993 American Association for Cancercancerres.aacrjournals.org Downloaded from

PATHOCENESIS OF MALIGNANT ASCITES

TA3/St MOT

ControlIC3Heb FeJi

1 d.»

3 day

4 day

7 d.y

to day

im/nj

Fig. 3. Influx and efflux of '"I-HSA into and oui of the peritoneal cavities at various intervals (0-10 days) after i.p. injection of TA3/St or MOT tumor cells. Tracer 125I-HSA wasinjected i.v. (Influx) or i.p. (Efflux). Data are expressed as either pi of plasma entering the peritoneal cavity (D) or as ul of peritoneal fluid leaving the peritoneum (•).The best-fit

lines were determined by linear regression.

regression. The slopes of each of the best-fit lines represent the cal- MOT-bearing mice (on days 7 and 10) and paralleled the observedculated rates of I25I-HSA tracer flow out of the plasma and into the increase in influx rates (Table 1; Fig. 4). By day 7 (TA3/St) and day

peritoneum (influx) or out of the peritoneal cavity and into the blood 10 (MOT) inflow had increased so as to exceed efflux, indicating the(efflux) (Table 1). point at which net ascites fluid accumulation would be expected. In

For both tumors, I25I-HSA tracer influx, as measured by the rate of fact, these data are in good agreement with the observed kinetics of

tracer appearance in the peritoneal fluid, remained unchanged (within peritoneal fluid accumulation (Fig. 2).experimental error) for several days after mice were given injections Rate Constants. The rate constants for 125I-HSA tracer transport

of either tumor and then rose dramatically between days 5 and 7 to were calculated from the experimentally determined flow rates andvalues 13- to 25-fold higher than control levels (Table 1; Fig. 4). On compartment volumes (see "Materials and Methods"). The values of

the other hand, I25I-HSA tracer efflux as measured either by the rate these kinetic parameters (*,, k2, and *6) changed with time followingof tracer disappearance from the peritoneal fluid or by the rate of tumor cell inoculation (Table 2; Fig. 4). The values of fc, and k2 fortracer appearance in the plasma decreased measurably (3- to 5-fold) '-3I-HSA in normal C-BSA mice and mice with fully developed

within 24 h after i.p. tumor cell injection, persisted at low levels for ascites were within experimental error of those determined earlier inseveral days, and then returned to normal or above-normal values the same strain of control and TA3/St or MOT ascites tumor-bearing(Table 1; Fig. 4). This resurgence of efflux was especially marked in mice using MT 70,000 F1TC-D as a tracer instead of I25I-HSA (24).

2634

Research. on October 17, 2020. © 1993 American Association for Cancercancerres.aacrjournals.org Downloaded from

Table 1

PATHOGENESISOF MALIGNANT ASCITES

tracer flov rates between peritoneal cavity ana plasma in ascites tumor-bearing and control mice"

InfluxDays

afteri.p.injectionoftumor

cellsControl

(A/J)TA3/S1-12567Rale

uttracerappearanceinperitoneal

fluid(Ml/min)''0.060.110.110.771.883.03Rate

oftracerdisappearancefrom

plasma(Ml/mm)*1.722.562.623.302.293.79EffluxRale

oftracerappearancein

plasma(Ml/min)''1.050.320.301.161.410.66Rale

ofIracerdisappearancefrom

perilonealfluid(ul/min)''4.560.941.072.432.592.27

C-BSA (A/J)'' 0.2(1 3.74 3.05 7.68

Control(C3Heb/FeJ)MOT'C-BSA

<C3Heb/FeJ)'/1347III0080.170.110.181.975.540.553.873.794.024.394.763.995.901.200.450.310.371.532.893.585.051.172.041.514.773.2012.25

" Flow rates are equivalent to the slopes of the best-fit lines as determined by linear regression analysis of the data in Fig. 3.'' The error in the kinetic parameters is estimated as ±30*. based on the error in the slopes of the regression lines in Fig. 3.' Ascites tumor-bearing animals received I x IO6 cells i.p. on day 0.'' To simulate ascitcs tumor peritoneal fluid volume and protein content, otherwise normal A/J and C3Heb/FeJ mice were given i.p. injections of 2 and 5 ml of 5% BSA, respectively.

Values for the kinetic parameter for peritoneal efflux into the plasma(A.-2)declined ~4- to 5-fold from control values within 24 h after i.p.

injection of either type of tumor cell. By day 7. As had decreased22-fold in TA3/St-bearing animals and. by day 10, 9-fold in MOT-

hearing mice. The values of kh, the rate constant for peritoneal effluxinto the extravascular space, declined 7- to 9-fold by day 1 in both

ascites tumors. In contrast to those efflux rate constants, values for therate constant for peritoneal influx (A:,) remained at normal controllevels (i.e., ~1 X 10~4 min"1) for 2 and 5 days after injection of

TA3/St and MOT tumor cells, respectively. Thereafter, k¡increaseddramatically in both tumor systems; i.e.. 23- and 46-fold in 7-dayTA.VSt and 10-day MOT-bearing animals, respectively. For both tu

mors, the values of the sum of k2 and £6(representing total efflux outof the peritoneal cavity) had decreased and the value oik, (representing influx into the peritoneal cavity) had increased such that they

became numerically equal (i.e., ~10 X 10 4 min ') by day 5 or 7

(Fig. 4); thereafter, the numerical values of A:,always exceeded thoseof the sum of k2 and kf, for both tumors (Fig. 4).

Measurement of 125I-HSA Influx and Calculation of Theoretical

Influx Curves Using a Three-Compartment Model. The percentage of the i.v.-injected dose of I25I-HSA measured in peritoneal fluid

and plasma at successive intervals after i.v. injection into tumor-bearing and control animals is plotted in Fig. 5. In control and C-BSA

A/J mice, a maximum of \% and 3%, respectively, of injected traceraccumulated in the peritoneal cavity at 180 min. Peritoneal traceraccumulation did not differ from that of control animals for up to fourdays after TA3/St tumor cell injection. However, by day 5 the amountof tracer present in the peritoneal fluid at 180 min had increasedmarkedly to 109Õ-of the injected dose; i.e.. 10 times that of controlmice and 3 times that of C-BSA mice. Thereafter, '2<iI-HSA accumu-

TA3/SI MOT

Fig. 4. A and B, I;M-HSA tracer flow rales (\i\l

min) into and out of the peritoneal cavity as afunction of time after i.p. inoculation of TA3/SI (A )and MOT (B) ascites tumor cells. Influx (O) isexpressed as the rate of tracer appearance in theperitoneal cavity following a bolus i.v. injection ofIracer: efflux is expressed as the rale of tracer appearance in the plasma (H) and as the rate of tracerdisappearance from the peritoneal cavity (A) following a bolus i.p. injection of tracer. C and /). rateconstants k\ (influx), k¡(efflux, rate of tracer appearance in the plasma), and the sum of the rateconstants ¿2and kt, (efflux, rale of tracer disappearance from the peritoneal cavity) for Iracer 12-SI-HSA

plotted as a function of time after i.p. inoculation ofTA.VSt (Ci and MOT (D) asciles tumor cells. •¿�AI; •¿�.fc;; A. k2 + *.>,.Bars, error in Ihe slopes ofthe best-fit lines shown in Fig. 3.

6-

<oX

10 i

Influx

Efflux-appearance

Efflux- disappearance6-

Influx

Efflux- appearance

Efflux- disappearance

10 12

6 8 024

Day* after ip tumor cell Injection

e io 12

2635

Research. on October 17, 2020. © 1993 American Association for Cancercancerres.aacrjournals.org Downloaded from

PATHOGENESIS OF MALIGNANT ASC1TES

Table 2 Kinetic parameters" k¡.k2. and k^for I251-HSA in control and ascites

tumor-bearing animals, and in otherwise normal animals (C-BSA) given

i.p. injections of 5% BSA to simulate tumor ascites fluid volumeand protein concentration

For comparison, values from an earlier study (24) using Mr 70,000 FITC-D as a tracerare included in parentheses. Values for rate constants are in units of min~l.

Control(A/J)TA3/SI'C-BSA

(A/J)JControl

(C3Heb/FeJ)MOT'C

BSA (C3Heb/FeJ)'/Days

after i.pinjection oftumorcells123567123457IIIIO4xk,h1

(0.5)11261423

(10)2

(2)(0.5)1146

(10)3

(2)IO4

xk2b44(50)97II)62

(2)14

(5)36(50)1077964

(1)7

(5)IO4

x*6*14617181156231171717272612016

" Values for the rate constants k,. k2. and *6 were calculated from the experimental

influx of efflux flow rales and the plasma and peritoneal volumes (see text).''The error in the kinetic parameters is estimated as ±50%.This estimate is based on

the error in the slopes of the lines in Fig. 3 (±30%)and the error in the volumes ofperitoneal fluid in Fig. 2 (±20%).

' Ascites tumor-bearing animals received 1 x IO6 cells ¡.p.on day 0.''To simulate ascites tumor peritoneal fluid volume and protein content, otherwise

normal A/J and C3Heb/FeJ mice were given i.p. injections of 2 and 5 ml of 5% BSA.respectively.

lation at 180 min ¡nthe peritoneum increased further to a maximumon day 6-7 that approximated 20% of the injected dose (20 times that

of control animals).Control C3Heb/FeJ animals behaved much like their A/J counter

parts. As much as 1% and 3% of injected tracer accumulated at 180min in the peritoneal cavities of control and C-BSA mice, respectively.Tracer accumulation in MOT-bearing C3Heb/FeJ mice approximated

that of control animals through day 4 (Fig. 5). However, by day 7,—¿�10% of the injected tracer appeared in the peritoneal cavities ofMOT-bearing animals by 180 min, and still greater peritoneal traceraccumulation was observed by day 10 (30% of injected dose, —¿�30

times that of control mice). Thus, the magnitude of increased traceraccumulation ultimately achieved in MOT-bearing C3Heb/FeJ micewas similar to that occurring in TA3/St-bearing A/J mice, although

these maximal values developed with somewhat different kinetics. Forboth tumor systems the peritoneal fluid accumulation maxima forI25I-HSA (occurring at 7 and 10 days for TA3/St and MOT ascites

tumors, respectively) closely approximated those we made previouslyfor the same tumors using M, 70,000 FITC-dextran as the tracer (24).

Using the experimentally determined rate constants (Table 2). thethree-compartment model depicted in Fig. 1, and the differential equa

tions characterizing that model (Appendix and Ref. 24). we generatedinflux and efflux curves describing I2SI-HSA tracer flux among the

several body compartments. These curves were then plotted alongwith the experimental influx and efflux data in Fig. 5. As can be seen,the theoretical curves provide an excellent fit of the experimental data.

Measurement of I25I-HSA Efflux and Calculation of Theoreti

cal Efflux Curves Using a Three-Compartment Model. The percentage of the i.p.-injected dose of I2ÃŽI-HSAtracer measured in the

peritoneal fluid and plasma of TA3/St and MOT ascites tumor-bearing

and control animals is also plotted in Fig. 5. In control A/J animals, amaximum of nearly 30% of injected tracer had entered the plasma by180 min. In TA3/St ascites tumor-bearing A/J mice, maximal values

for the percentage of tracer accumulating in plasma were markedly

reduced as a function of time following tumor cell inoculation: 20%of the injected dose on day 1; 10% on days 2-5; 5% on day 6: and only3% on day 7; reflecting a 10-fold decline compared with controlanimals. The corresponding value for I2ÕI-HSAaccumulation in C-

BSA mice was 20%.A similar pattern was observed in control and MOT tumor-bearing

C3Heb/FeJ mice. In control animals, —¿�40%of i.p.-injected tracer

accumulated in plasma at 180 min. On days 2 through 4 after MOTtumor cell injection. C3Heb/FeJ mice achieved maximum plasmalevels of only 10% of injected tracer dose; by day 7 this value haddecreased to 7% and by day 10 to 5%. In C-BSA mice, the corre

sponding value was 10% of injected dose.As was the case for influx studies, the efflux results with 125I-HSA

tracer for 7-day TA3/St and 10-day MOT ascites tumor-bearing mice

are in good agreement with our published values using MT 70,000FITC-D as a tracer (24). Similarly, the three-compartment model was

able to predict efflux curves that closely matched the experimentaldata for both the plasma and peritoneal compartments in ascites tumor-bearing as well as in control animals (Fig. 5).

Influx and Efflux Studies with "Cr-RBC as Tracer

Flow Rates. At later stages of tumor growth, tumor ascites oftenbecame bloody. However, at early intervals (up to day 7 or 10 inTA3/St- and MOT-bearing mice, respectively) and in control andC-BSA animals, <0.1% of i.v. injected 51Cr-RBC entered the perito

neal cavity within 180 min; i.e., the influx rate was essentially zero(data not shown). In addition, <0.5% of either i.v. or i.p.-injected'''Cr-RBC was taken up by the spleen at intervals of 5 to 360 min

following tracer injection; i.e., the short-term clearance by the spleen

was essentially zero (data not shown).Following i.p. injection. 5'Cr-RBC readily exited the peritoneal

cavity of control animals. Data for 5'Cr-RBC tracer efflux (as meas

ured both by the appearance of tracer in the blood and by the disappearance of tracer from the peritoneal cavity) are plotted as a functionof time at different intervals following tumor cell inoculation (Fig. 6).Flow rates are taken as the slopes of the best-fit lines as determined by

linear regression (Table 3).The pattern of efflux of 5ICr-RBC from the peritoneal cavity was

similar in normal control mice of both strains. As measured by the rateof tracer disappearance from the peritoneal cavity, efflux decreasedsubstantially following the i.p. injection of either tumor cell line; i.e.,2.0-fold for TA3/St by day 1 and 1.6-fold for MOT by day 2. Although

reported here only through day 3 (Table 3). these low levels persistedthroughout the entire course of ascites tumor growth. As measured bythe rate of tracer appearance in the blood, efflux decreased even moremarkedly (i.e., 8- to 12-fold by 24 h) for both tumor systems; this

decrease also persisted throughout the observation period (Table 3)and beyond.

Measurement of 5lCr-RBC Efflux and Calculation of Theoret

ical Efflux Curves Using a Three-Compartment Model. The percentage of the i.p.-injected dose of MCr-RBC tracer measured in the

peritoneal fluid and blood of TA3/St and MOT ascites tumor-bearingand control animals is plotted in Fig. 7. In normal control and C-BSAA/J mice, respectively, —¿�10-20%and —¿�5-10%of the injected dosesof "Cr-RBC were detected in the blood by 180 min; these values for

the accumulation maxima are not much lower than those observedwhen 125I-HSA was used as a tracer in control and C-BSA A/J mice(Fig. 5). However, the percentage of injected MCr-RBC that left the

peritoneum and entered the blood declined detectably as early as 6 hafter i.p. TA3/St tumor cell inoculation and was even more pronounced by 24 h (Figs. 7 and 8). Efflux of MCr-RBC was detectably

impaired in C3Heb/FeJ mice as early as day 1 after i.p. MOT tumorcell injection; at this time, —¿�5%of the injected dose ofs'Cr-RBC was

detected in the blood by 180 min. Efflux was impaired substantially

2636

Research. on October 17, 2020. © 1993 American Association for Cancercancerres.aacrjournals.org Downloaded from

PATHOCENESIS OK MALIGNANT ASCITES

TA3/SI MOT

7 (lay

7 a,y

10 d«y

Fig. 5. Semilog plots describing ihe influx of I25I-HSA from ihe plasma to the peritoneal cavity at various intervals after i.V. tracer injection and the efflux of '"M-HSA from the

peritoneal cavity to the plasma at various intervals after i.p. tracer injection in normal mice, C-BSA mice, and mice bearing TA3/St and MOT ascites tumors for number of days indicated.I2^I-HSA concentration (iinltuuì?}is expressed as the percentage of injected dose present in either the plasma or in the peritoneal fluid as a function of time (tih.tcisxa) following tracerinjection. Data from at least one and up to 5 animals are included at each time interval after tracer injection. Cun-ea represent the computer-generated influx and efflux curves, usingthe three-compartment model (Fig. I and Appendix) and the experimentally determined kinetic parameters presented in Table 2. These rate constants, together with the three-

compartment model depicted in Fig. 1. and the differential equations that describe that model, were used to generate the theoretical influx and efflux curves. In these calculations weset the values for the non-experimentally determined rate constants as follows: k3 = 0.002; In = 0.1X12;k> = 0.003; based on literature values (36). D. plasma; •¿�.peritoneal fluid.

further by day 3 when only I% of the i.p.-injected dose of "Cr-RBC

appeared in the blood at 180 min, almost a 10-fold decrease relative

to control levels in normal C3Heb/FeJ mice.The rate constants for "Cr-RBC tracer efflux calculated from the

experimentally determined flow rates and compartment volumes (see"Materials and Methods") are summarized in Table 4. Values for k:,

the rate constant for tracer appearance in the blood, decreased substantially in ascites tumor-bearing animals relative to those of their

normal counterparts. By 1 day after tumor cell injection, k2 haddecreased 5- and 11-fold in MOT- and TA3/St-bearing animals, re

spectively. The rate constant for tracer appearance in the peritonealinterstitium, A:6.was found to be quite low in control animals, relativeto that of '"I-HSA and, in contrast to I25I-HSA. underwent little

change in ascites tumor-bearing animals (compare Tables 2 and 4).The "Cr-RBC efflux rate constants (Table 4) and the differential

equations that describe the three-compartment model (24) were used

to determine the efflux curves that characterized tracer transport. In

these calculations, the values of kt, A:,,A:4,and kf were set to equal zerobecause insignificant numbers of circulating "Cr-RBC entered the

peritoneal cavity, spleen, extravascular space of other tissues, or theurine during the course of these experiments in any group of animals.As with '^I-HSA as the tracer, the generated curves compared favor

ably to the experimental values for percentage of injected dose of"Cr-RBC tracer found in the blood or retained in the peritoneal cavity

at various time intervals after i.p. tracer injection (Fig. 7).Comparison of I25I-HSA and *'Cr-RBC Efflux. Efflux rates for

l25I-HSAand s'Cr-RBC in normal. C-BSA, and ascites tumor-bearing

mice are compared in Fig. 8. In normal control animals, the rate ofappearance in the blood of i.p.-injected "Cr-RBC was nearly equal tothe rate of appearance in the plasma of i.p.-injected '-SI-HSA (Tables

1 and 3; Fig. 8, A and B). In C-BSA animals, both tracers entered thevascular space at enhanced (~2- to 3-fold) rates (Fig 8, A and fi). In

mice bearing either ascites tumor, the rate of appearance of bothtracers in the blood (or plasma) declined as tumor growth progressed.

2637

Research. on October 17, 2020. © 1993 American Association for Cancercancerres.aacrjournals.org Downloaded from

PATHOOF.NESIS OF MALIGNANT ASCITF.S

5 1

Fig. 6. Efflux of ''Cr-RBC from the peritoneal

cavity as a function of time following i.p. injectionof tracer in control, C-BSA, and TA3/SI- and MOT-

bearing mice. Disappearance of tracer from theperitoneal cavity and appearance of tracer in theblood were measured separately. Data are presentedas the volume (fil) of peritoneal Huid entering theblood (D) or as the volume (ul) of peritoneal Huidleaving the peritoneal cavity (•)at each time point.The best-fit lines were determined by linear regres

sion.

_^y = 10936*11938» R'2 - 0 885

Cr-RBC

300<

200

100 •¿�y - - 8 9893 + O 7054BI R*2 - O 753

Õ"B"

Control(C3H.6/F.J!

t d.y

60 120 180 240 300

500

\ day

400 •¿�

300 •¿�

200

100

y - 37519 - 0549461 R"? . 0429

y = 3 2749 t 4 3117«-2» R*2 - 0 277D

2 day

3 day 3 d«y

60 120 180 240 300 time (min) 60 120 180 240 300

but that of 51Cr-RBC fell more precipitously than that of I25I-HSA;i.e., a -90% fall in the rate of appearance of s'Cr-RBC in the blood

by day 1 as compared with a ~70% decline for the rate of appearanceof 125I-HSA in the plasma by day 3. Thus, in ascites tumor-bearinganimals, the efflux of 51Cr-RBC from the peritoneal cavity, as meas

ured by the rate of tracer appearance in the blood, was impairedearlier and more substantially than that of 125I-HSA.

Whereas the rates of appearance in the vascular space on i.p.-injected I25I-HSA and 5'Cr-RBC were nearly equivalent in normal

mice, the rates of disappearance of these tracers from the peritonealcavity differed significantly even in normal mice; i.e., the disappearance rate of I25I-HSA exceeded that of 5ICr-RBC in normal mice by

a factor of 3- to 6-fold (Fig. 8, C and D; Tables 1 and 3). Similarly, inascites tumor-bearing mice, the rates of disappearance of both tracers

from the peritoneum declined as a function of tumor cell growth, ashad the rates of appearance of these tracers in the vascular space.However, as in normal animals, the rate of disappearance of I25I-HSAalways exceeded that of 5lCr-RBC by a substantial margin (1.25- to

4-fold).

DISCUSSION

We have investigated the nature and kinetics of the disrupted equilibria that result in ascites fluid accumulation in two syngeneic mousetumor models. Our results provide the first quantitative description ofthe early changes in transperitoneal transport that are responsible forthe development of tumor ascites. In normal mice, peritoneal fluid iscontinuously generated and resorbed, but, because influx is relativelylow and efflux relatively high, résorptiongreatly exceeds influx (57),and, at steady-state, very little fluid is found in the peritoneum (~0.1

ml and 0.45 ml in normal A/J and C3Heb/FeJ mice, respectively).However, within 24 h of i.p. injection of either tumor cell line,transport out of the peritoneal cavity declined dramatically for both125I-HSA and 5lCr-RBC (Figs. 3, 4, 6, and 8). whereas influx rates did

not change appreciably, i.e., remained within experimental error ofcontrol values for I25I-HSA (Figs. 3 and 4) and essentially at zero for51Cr-RBC. Only when the inflow rate for 125I-HSA had increased

significantly (5- to 10-fold by 5-7 days after i.p. tumor cell injection)

to exceed outflow did we observe net ascites fluid accumulation (Figs.

2638

Research. on October 17, 2020. © 1993 American Association for Cancercancerres.aacrjournals.org Downloaded from

PATHíXil-.NhSIS Oí MALIGNANT ASCITfcS

Table 3 ^'Cr-RBC tracer efflux flow rates between the peritoneal cavity and bloodin ascites tumor-bearing and control mice"

Control(A/J)TA3/SfDays

after i.p.injection oftumorcells0.2513Rate

oftracerappearance

in blood(ul/min)h1.190.600.100.02Rate

oftracerdisappearance

from peritonea]Huid (ni/min)''1.511

(K>0.730.42

C-BSA <A/Jr

Control (C3Heb/FeJ>MOT''

C-BSA(C3Heb/FeJ)

2.01

0.710.090.040.02

1.92

5.82

0.860.780.550.52

3.88" Row rales are equivalent to the slopes of the best-tit lines as determined by linear

regression analysis of the data in Fig. 6.'' The error in the kinetic parameters is estimated as ±30%.based on the error in the

slopes of the lines in Fig. ft.' Asches tumor-bearing animals received I x \0h cells i.p. on day 0.''To simulate ascites tumor peritoneal fluid volume and protein content, otherwise

normal A/J and C3Heb/FeJ mice were given i.p. injections of 2 and 5 ml of 5r/íBSA.

respectively.

2 and 4). Thus, impaired peritoneal outflow of the type and extentinduced by i.p. tumor cell injection was not sufficient by itself to causeperitoneal fluid accumulation. This conclusion is consistent with earlier work showing that abrasion injury to the diaphragm, which decreased efflux by an order of magnitude, did not by itself induceperitoneal fluid accumulation (58. 59). Whether impaired peritonealefflux is nonetheless necessary (although by itself insufficient) forperitoneal fluid accumulation, or whether increased influx alone caninduce ascites, was not addressed by our experiments and requiresinvestigation. Patients with peritoneal carcinomatosis (4, 12-16) also

exhibit decreased peritoneal absorption when compared to those withnonmalignant ascites (e.g., induced by cirrhosis; Refs. 15 and 38) orchronic peritoneal dialysis (38. 60).

We have no satisfactory explanation for the early and dramaticdecline in the efflux of macromolecular tracers that developed withinhours of i.p. tumor cell injection. This change clearly preceded anyincrease in peritoneal fluid volume or protein content; moreover,increases in peritoneal fluid and protein would be expected to favorefflux (probably by increasing intraperitoneal pressure; Refs. 25, 26,61, 62), as was the case in C-BSA mice of either strain (Tables I and

3; Fig. 8). Histológica! studies (not shown) revealed no anatomicalbasis for impeded peritoneal outflow (e.g., plugging of lymphatic-

channels by tumor or inflammatory cells) such as have been describedat later stages of ascites tumor growth (12, 13, 17, 63). It will be ofinterest to determine whether tumor cells or tumor cell product(s)close the diaphragmatic lymphatics or their stornata by a functionalmechanism.

Particles such as 5lCr-RBC are reported to exit the peritoneal cavity

exclusively by way of draining lymphatics which are concentrated inthe diaphragm (18, 21, 23, 36, 38^-0), whereas soluble proteins suchas 125I-HSA may also exit the peritoneum by additional routes; e.g.,

passage into the parietal or visceral interstitium (23) followed byuptake via abdominal visceral lymphatics (22. 64) or direct entry intoperitoneal wall capillaries (65). If these assumptions are correct, thenthe rate of peritoneal efflux of i.p.-injected 5'Cr-RBC should provide

a reliable estimate of peritoneal cavity lymphatic drainage. Moreover,equivalent results should be obtained by measuring either disappearance from the peritoneal cavity of i.p.-injected tracers or their appearance in the blood.5

5 The validity of estimating peritoneal lymphatic flow rates by either of these meas

urement is based on two assumptions: («)intraperitoneal fluid is drained by the peritoneal

In our experiments, the rate of appearance of i.p.-injected 51Cr-RBC

in the blood approached (i.e., achieved 79-83% of) its rate of disap

pearance from the peritoneal cavity in normal mice of both strains(Table 3). allowing us to consider 51Cr-RBC appearance in the blood

as an approximation of the lymphatic drainage rate in normal mice(i.e., 1.0 ul/min). However, in ascites tumor-bearing mice the rates ofMCr-RBC appearance in the blood were not equivalent to the rates of

tracer disappearance from the peritoneal cavity; only 4-13% of the5lCr-RBC leaving the peritoneum appeared in the blood. One explanation for this finding is that "Cr-RBC (or 5lCr liberated from RBC)

leave the peritoneal cavities of ascites tumor-bearing animals not only

by lymphatics but also by other mechanisms, e.g., binding to thealtered peritoneal surface, phagocytosis and degradation by macrophages or other inflammatory cells present in the peritoneal wall ofsuch animals, etc. Also, i.p.-injected 5ICr-RBC might become altered

in ascites tumor-bearing mice such that they undergo more rapid

clearance by the reticuloendothelial system after returning to the bloodfrom the peritoneal cavity. Finally, tumor ascites fluid accumulation isalso associated with injury to mesothelial cells, angiogenesis, fibrosis,etc. (29), changes which may be expected to alter properties of peritoneal permeability to tracer and tracer flow rates. Whatever themechanism(s) responsible, the divergence between peritoneal disappearance and blood appearance rates of i.p.-injected 5lCr-RBC indi

cate that this tracer does not provide an accurate measure of lymphaticdrainage rates in ascites tumor-bearing animals.

Whereas efflux rates of "Cr-RBC remained depressed for as long

as animals were followed (7 days for TA3/St and 25 days for MOT,data not shown), efflux rates of I25I-HSA, whether measured by

disappearance from the peritoneal cavity or by appearance in theplasma, returned (at least transiently) from their nadirs to normal oreven to above-normal values in both tumor systems (after day 2 inTA3/St-bearing mice, after day 4 in MOT-bearing mice) (Table I ; Fig.4). This transient increase in I2SI-HSA efflux flow rates correlates

with increasing accumulation of ascites fluid which, among othereffects, causes increased hydrostatic pressure and therefore probablyleads to stretching of the diaphragmatic lymphatics so as to maintaintheir stornata in the open position (25. 26. 61, 62). Indeed, this effectwas mimicked in C-BSA mice with "artificial ascites" (Fig. 8); how

ever, other as yet unexplored possibilities also exist; e.g., the openingof additional preexisting tissue lymphatics and the induction of newlymphatics.

Fluid and plasma proteins normally enter the peritoneum from themicrovessels that line the peritoneal cavity. The fluid that began toaccumulate in the peritoneal cavity ~4 days after i.p. tumor cell

injection differed from that found in normal animals in composition aswell as in volume. Although published values for the protein concentration of normal peritoneal fluid differ widely within and acrossspecies (range of 18-25 mg/ml in the rat. 25 mg/ml in the rabbit.32-57 mg/ml in the cat, and 42 mg/ml for humans) (30, 66-70), our

data indicate that in normal mice this fluid has the properties of atransudate with a protein content ~20-25% that of plasma (i.e., 10-15mg/ml). By contrast, the protein content of tumor ascites was sever-alfold greater than that of normal animals, eventually approaching—¿�85%of plasma levels; moreover, albumin was a prominent compo

nent of tumor ascites fluid and accumulated in a proportion similar tothat found in plasma. This last finding indicates that the substantiallyincreased protein content of tumor ascites cannot be attributed to the

lymphatics without an increase or decrease in concentration; and (hi the intraperitonealtracers used in such measurements are returned to the venous circulation exclusively bythe peritoneal lymphatics and in a form that does not alter their subsequent blood clearancerate (38). If both of these criteria are met. then the rate of tracer appearance in the bloodshould equal its rate of disappearance from the peritonea] cavity and lymphatic absorptioncan be assessed accurately by either measurement.

2639

Research. on October 17, 2020. © 1993 American Association for Cancercancerres.aacrjournals.org Downloaded from

PATHOGENESIS OF MALIGNANT ASCITES

TA3/S1Efflux

Fig. 7. Semilog plots describing the efflux of' 'Cr-RBC from the peritoneal cavity to the blood at

various intervals (ithscissa ) after i.p. injection innormal mice. C-BSA mice, and in mice bearing

TA.VSt and MOT ascites tumors for the number ofdays indicated. 5'Cr-RBC concentration (ordinale)

is expressed as the percentage of injected dosepresent in either the blood or in the peritoneal fluid.Data from at least one animal and from up to 5animals are included at each time interval aftertracer injection. Cun'f.v represent the computer-generated efflux curves, using the three-compartment model (Fig. I) and the kinetic parameterspresented in Table 4. These rate constants, togetherwith the three-compartment model depicted in Fig.

1. and the differential equations that describe thatmodel, were used to generate the theoretical influxand efflux curves. In these calculations we set thevalues for the other rate constants. k\. ki. k¿.and £>.all equal to zero. D, blood; •¿�.peritoneal fluid.

8 2

Cr-RBC

Control(A/J)

6 hr

i day

3 day

Control(C3Heb FeJ

i day

2 day

3 day

C-BSA

time (min)60 120

local release of tumor proteins, as the result of either tumor cellsecretion or death. Rather, it reflects increased permeability of themicrovessels lining the peritoneal cavity to plasma proteins. Thus,whereas the small amount of peritoneal fluid present in normalanimals is a transudate, that which accumulates in tumor ascites isan exúdate,an interpretation also favored by some earlier workers(1-4*7,8).

While the mechanism(s) that impede peritoneal efflux within hoursof tumor cell injection remain obscure, we can propose a likely explanation for the increased tracer influx and high protein content ofperitoneal fluid that developed subsequently. Studies from our laboratory have shown that a wide variety of tumor cells, including thosestudied here, produce and secrete VPF, a Mr 34,000-42,000, disulfide-bonded dimeric protein which, in low nanomolar to picomolar concentrations and with a potency some 50,000 times that of histamine,dramatically increases the permeability of venules and small veins toplasma proteins (8). Abundant VPF activity has been found in ascitesfluid (8, 71), and we have recently demonstrated its localization in theperitoneal wall vessels of guinea pigs and mice bearing ascites tumors(29, 72, 73). Moreover, in a syngeneic guinea pig ascites tumorsystem, Yeo et al. (71) have recently found that the amount of VPF in

peritoneal fluid correlates with the volume of fluid accumulation andfollows kinetics similar to those of the two mouse ascites tumorsstudied here. Taken together, these results strongly suggest that VPFis responsible for the increased influx of fluid and plasma proteins thatcharacterize both of our mouse ascites tumors after about day 4.Therefore, the delayed onset of increased fluid influx characteristic ofour tumor models may reflect (a) a lag phase required for VPFexpression (synthesis or secretion) by peritoneal tumor cells; (b) theneed to accumulate a sufficient minimal concentration of VPF forenhanced peritoneal vessel permeability; (r) an alteration of the integrity of the mesothelial lining so that VPF can diffuse from theperitoneal cavity to target vessels in the peritoneal wall; or (d) theneed for angiogenesis in the peritoneal lining to provide sufficientnumbers of blood vessels to support increased influx. The third ofthese explanations seems unlikely, if, as has been proposed (22, 23,32, 33), plasma proteins, all of which are larger than VPF, normallyescape from the peritoneum by crossing an intact mesothelium. Thelast explanation also seems unlikely in that influx can be at leasttransiently increased in normal guinea pigs by i.p. injections of VPF(8). Further experiments will be required to sort out these and otherissues.

2640

Research. on October 17, 2020. © 1993 American Association for Cancercancerres.aacrjournals.org Downloaded from

PATHOGF.NESIS OF MALIGNANT ASC1TES

Fig. 8. Comparison of the rates at which i.p.-injected I2*I-HSA (•)and "Cr-RBC (&) appear in

plasma or blood, respectively (A and B) or disappearfrom the peritoneal cavity (C and /> I in normal mice,C-BSA mice, and mice hearing either TA3/St (A andC) or MOT (B and Dìascile* tumors.

TA3/St

3 -

2 -

1251-HSA

51Cr-RBC

iControl 0.25 1

16

umÒeì 8 H

'liIH

1251-HSA

51Cr-RBC

iControl 0.25 1

C-BSA

C-BSADays after Ip tumor cell inoculation

MOT

3 -

2 -

1251-HSA

51Cr-RBC

16

12 -

8 -

4 -

O.

1251-HSA

51Cr-RBC

1 I

n r T i iControl 1237 C-BSA

Control 1237 C-BSADays after ip tumor cefi inoculation

Tahle 4 Kinetic parameters" k¿and kf,for ^'Cr-RBC in control and ascites

tumor-hearing animals, and in otherwise normal animals given i.p. injections of5ck BSA to simulate tumor ascites fluid volume and protein concentration

Values for rate constants are in units of min"'.

Control(A/J)TA3/S1'C-BSA

(A/J)''Control

(C3Heh/FeJlMOT'C-BSA

(C3Heb/FeJ)''Days

after ¡.p.injection oftumorcells0.2513123\0>xk2h22II218II210.53IO4 xk„699615210773

" Values for the rate constants fcj and £<,were determined from the experimental efflux

tlow rates and the blood and peritoneal fluid volumes (see text).'' The error in the kinetic parameter is estimated as ±50^. This estimate is based on the

error in the slope of the lines in Fig. 6 (±30<£) and the error in the volumes of peritoneal

fluid in Fig. 2 (±20%).' Ascites tumor-bearing animals received I x I0ft cells i.p. on day 0.f/To simulate ascites tumor peritoneal fluid volume and protein content, otherwise

normal A/J and C3Heb/FeJ mice received 2 and 5 mol of 5% BSA i.p., respectively.

APPENDIX: PHARMACOKINETIC MODEL

Influx into the peritoneum requires that circulating macromolecules passthrough a succession of barriers that include the vessel wall and its underlyingbasal lamina, interstitial connective tissue, and the mesothelium (reviewed inRet. 74). Efflux from the peritoneal cavity into the blood requires that molecules either drain via lymphatic channels concentrated in the diaphragm orreenter the blood directly by retracing the steps summarized above for influx(reviewed in Refs. 74. 75). Influx and efflux are complex, multifaceted processes, depending at once on properties of the tracer molecule (e.g., size shape,charge), on host factors (vascular permeability, hydraulic conductivity, andreflectivity, peritoneal volume, composition, and pressure), as well as on theprinciples of convection and diffusion (74. 75). While a complete understanding of macromolecular transfer will require the measurement of each of theseparameters, it is useful, as a first approximation, to consider influx and effluxas composite entities, representing the summation of all of the variables listedabove. For this purpose we chose a simplified mathematical model (49) composed of three compartments: the intravascular space: the peritoneal cavity:

and all extravascular spaces other than the peritoneal cavity (Fig. 1). Thismodel makes no attempt to measure any specific physiological process. Rather,all factors that modify transport between compartments are lumped together inrate constants (e.g.. k¡,k2, etc.) that do not distinguish among such componentsas diffusion or convection, vascular permeability, vascular area. etc.

The macromolecular transport rate constants calculated from this modelcompare favorably with with those previously reported in the literature. Thecalculated values we obtained for the rate constants it, and k2 with |:<1I-HSA as

tracer are similar to those we reported earlier (24) for comparably sizedFITC-D tracers in normal mice and in mice with fully developed tumor ascites

(Table 2). There have been relatively few other measurements of macromolecular influx into the peritoneum in normal animals (35. 76-78). Flessner etal. (35) found that ~3% of an i.v.-injected dose of I25I-BSA appeared in theperitoneal cavity at 180 min; a value for k, of ~2 x 10~4 min"1 may be

inferred from these influx data. These results are in general agreement with ourown. although the Flessner study was performed on anesthetized animals anddata were analyzed with a different distribution model.

Conversely, a number of studies exist concerning the efflux of macromolecules from the peritoneal cavity. Our model-fitted values for k¿in normalanimals (36 and 44 x IQ-4 min"' for 125I-HSA or 11 and 22 X IO-" min'1 for

5'Cr-RBC) are in close agreement with values either reported or inferred from

earlier analyses of peritoneal efflux in several species of normal animals witha variety of tracers, e.g.. 20-58 x IO"4 min"' (18. 58, 66, 79). Values toward

the lower end of this range probably reflect the fact that experiments wereperformed on anesthetized animals. Lymphatic absorption depends heavily ondiaphragmatic contractions which are reduced by general anesthetics (19): e.g.,studies with variably sized methylacrylamide copolymer tracers in normalunanesthetized rats demonstrated that 407r of an i.p.-injected dose appeared inthe blood by 90 min (yielding a value tor k2 of ~44 X 10~4 min"'), whereas

only I29i of the i.p. administered dose appeared in the blood by 90 min inetherized animals (yielding a value of As of ~13 X IO"4 min"1), and only 2%

of the dose was transferred to the plasma when the animals were anesthetizedwith fluothane (yielding a value of k2 of 2 X IO"4 min"1) (54). Our results are

also in agreement with values for k? interred from data for the efflux of tracer5lCr-RBC from the peritoneal cavities of normal animals (42 X IO"4 min"')and animals with incipient tumor ascites (24 and 2 x IO"4 min"' on days 1 and

4, respectively, after i.p. tumor cell injection (12). Similarly. Sträube (7)obtained a value tor k2 of 4 X IO"4 min"' on day 3 after i.p. inoculation of

another murine ascites tumor.To our knowledge, no other "black-box" model of transperitoneal transport

has included a specific rate constant for peritoneum-to-extravascular space

transport (our it,,), although other investigators have recognized the contribution of interstitial absorption to peritoneal drainage in "distributed" models of

2641

Research. on October 17, 2020. © 1993 American Association for Cancercancerres.aacrjournals.org Downloaded from

PAimxil.M.SIS 01 MAI.KÕNAMASCITI*

35.

36.

37.

transport kinetics (35. 80). Initially, we attempted to model our influx andefflux data without using the rate constant kh. but modeling (particularly of theefflux data) was less successful than that shown here.

REFERENCES

1. I i i.ii.i J.. and Dumont. A. E. Paihogenesis of ascites in mice with peritoneal carci-nomatosis. J. Nati. Cancer Inst.. 56. 547-549. 1976. 38.

2. Garrison. R. N.. Kaelin. L. D.. Heuser. L. S.. and Galloway. R. H. Malignant ascites.Clinical and experimental observations. Ann. Surg.. 203: 644-649. 1986. 39.

3. Garrison. R. N.. Galloway, R. H.. and Heuser. L. S. Mechanisms of malignant ascitesproduction. J. Surg. Res.. 42: 126-132. 1987.

4. Hirabayashi. K.. and Graham. J. Genesis of ascites in ovarian cancer. Am. J. Obstet. 40.Gynecol.. 106: 492-497, 1970.

5. Lacy. J. H.. Wieman. T. J.. andShively. E. H. Management of malignant ascites. Surg. 41.Gynecol. Obstet.. 159: 397-412. 1984.

6. Runyon. B. A., Hoefs. J. C. and Morgan. T. R. Ascine fluid analysis in malignancy-related ascites. Hepatology. H: 1104-1109. 1988. 42.

7. Sträube, R. L. Fluid accumulation during initial stages of ascites tumor growth.Cancer Res.. 18: 57-65, 1958.

8. Senger. D. R.. Galli. S. J.. Dvorak. A. M.. Peruzzi, C. A.. Harvey. V. S.. and Dvorak. 43.H. F. Tumor cells secrete a vascular permeability factor that promotes accumulationof ascites fluid. Science (Washington DC), 21V: 983-985. 1983.

9. Senger. D. R.. Peruzzi, C. A.. Feder. J.. and Dvorak. H. F. A highly conserved vascular 44.permeability factor secreted by a variety of human and rodent tumor cell lines. CancerRes.. 46: 5629-5632. 1986.

10. Senger, D. R.. Connolly, D. T.. Van De Water. L.. Feder, J.. and Dvorak. H. F. 45.Purification und NHi-temiinal amino acid sequence of guinea pig tumor-secretedvascular permeability factor. Cancer Res., 50: 1774-1778. 1990. 46.

11. Dvorak. H. F.. Nagy. J. A.. Dvorak, J. T., and Dvorak, A. M. Identification andcharacterization of the blood vessels of solid tumors that are leaky to circulating 47.macromolecules. Am. J. PathoL 133: 95-109. 1988.

12. Feldman. G. B.. Knapp. R. C., Order. S. E.. and Hellman. S. The role of lymphaticobstruction in the formation of ascites in a murine ovarian carcinoma. Cancer Res.. 48.32: 1663-1666. 1972.

13. Feldman. G. B. Lymphatic obstruction in carcinomatous ascites. Cancer Res., .15: 49.325-332. 1975.

14. Feldman, G. B.. and Knapp, R. C. Lymphatic drainage of the peritoneal cavity and its 50.significance in ovarian cancer. Am. J. Obstet. Gynecol., 119: 991-994, 1974.

15. Bronskill, M. J.. Bush, R. S.. and Ege, G. N. A quantitative measurement ot peritoneal 51.drainage in malignant ascites. Cancer (Phila.l. 40: 2375-2380. 1977.

16. Coates, G.. Bush, R. S., and Aspin. N. A study of ascites using lymphoscintigraphy 52.with ""re-sulfur colloid. Radiology, 107: 577-583. 1973.

17. Pavelic, Z., Boranic, M., Pavelic. K.. and Vasarevic. M. Pathogenesis of ascites in a 53.murine transplantable reticulosarcoma type A. Z. Krebsforsch.. SS: 9I-95. 1976.

18. McKay, T.. Zink. J., and Greenway. C. V. Relative rates of absorption of fluid and 54.protein from the peritoneal cavity in cats. Lymphology. //: 106-110, 1978.

19. Morris. B. The effect of diaphragmatic movement on the absorption of protein and ofred cells from the peritoneal cavity. Ausi. J. Exp. Biol., 31: 239-246, 1953.

20. Holm-Nielsen. P. Pathogenesis of ascites in peritoneal carcinomatosis. Acta PathoL 55.Microbiol. Scand., 33: 10-21, 1953.

21. Courtice. F. C., Harding, J., and Steinbeck, A. W. The removal of free red blood cells 56.from the peritoneal cavity of animals. Aust. J. Exp. Biol., 31: 215-225, 1953.

22. Abemethy. N. J.. Chin, W.. Hay, J. B., Rodela, H.. Oreopoulos. D.. and Johnston. M. 57.G. Lymphatic drainage of the peritoneal cavity in sheep. Am. J. Physiol.. 260.F353-F358. 1991. 58.

23. Flessner, M. F.. Parker, R. J., and Sieber, S. M. Peritoneal lymphatic uptake offibrinogen and erythrocytes in the rat. Am. J. Physiol.. 244: H89-H96. 1983.

24. Nagy. J. A., Herzberg, K. T., Masse, E. M., Zientara. G. P.. and Dvorak. H. F 59.Exchange of macromolecules between plasma and peritoneal cavity in ascites tumor-bearing, normal, and serotonin-injected mice. Cancer Res.. 49: 5448-5458, 1989. 60.

25. Zink, J., and Greenway, C. V. Intraperitoneal pressure in formation and reabsorptionof ascites in cats. Am. J. Physiol.. 233: HI85-HI90. 1977.

26. Zink. J., and Greenway. C. V. Control of ascites absorption in anesthetized cats: 61.effects of intraperitoneal pressure, protein, and furosemide diuresis. Gastroenterology,73: 1119-1124, 1977.

27. Mattes, M. J. Biodistribution of antibodies after intraperitoneal or intravenous injec- 62.lion and effect of carbohydrate modifications. J. Nati. Cancer Inst., 79; 855-863,

1987. 63.28. Nagy. J. A.. Herzberg. K. T.. and Dvorak, H. F. Stroma generation in ascites tumors.

J. Cell Biol., 109: 180a. 1989.29. Nagy. J. A.. Masse, E. M.. Meyers, M. S., and Dvorak, H. F Peritoneal angiogenesis 64.

and desmoplasia accompany ascites tumor growth. Thromb. Haemostasis, 65: 1051,1991.

30. Markowitz, H. Proteins in ascites and ascites fluid. In: Proteins in Body Fluids. 65.Amino Acids, and Tumor Markers: Diagnostic and Clinical Aspects, pp. 221-226.

New York: Alan R. Liss. 1983.31. Lifshitz. S., Clarke, W., and Buchsbaum. H. J. Serum and ascitic fluid protein 66.

electrophoresis and immunoglobulin levels in gynecologic cancer. Gynecol. Oncol.,S: 31-36, 1980. 67.

32. Ressner, M. F, Fenslermacher, J. D., Blasberg, R. G., and Dedrick, R. L. Peritonealabsorption of macromolecules studied by quantitiative autoradiography. Am. J. Physiol.. 248: H26-H32. 1985. 68.

33. Ressner. M. L.. Dedrick. R. L.. and Reynolds, J. C. Bidirectional peritoneal transportof immunoglobulin in rats: tissue concentration profiles. Am. J. Physiol., 26.?: F15- 69.

F23. 1992.34. Leypoldl. J. K.. Chiù.A. S.. Frigon. R. P., and Henderson, L. W. Dialysate to blood

2642

transport of macromolecules during peritoneal dialysis. Am. J. Physiol.. 257: HI 851-HI859. 1989.Flessner. M. F. Dedrick. R. L.. and Schultz. J. S. Exchange of macromoleculesbetween peritoneal cavity and plasma. Am. J. Physiol.. 248: HI5-H25. 1985.Bettendorf. U. Lymph How mechanism of the subperitoneal diaphragmatic lymphatics. Lymphology. //: 111-116. 1978.

Allen. L. On the penetrability of the lymphatics of the diaphragm. Anat. Ree., 124:639-657. 1956.

Mactier. R. A.. Khanna. R.. Twardowski, Z. J.. and Nolph. K. D. Role of peritonealcavity lymphatic absorption in peritoneal dialysis. Kidney Int.. 32: 165-172. 1987.Yoffey. J. M.. and Counice. F. C. Lymph flow from regional lymphatics. In: Lymphatics, Lymph and the Lymphomyeloid Complex, pp. 206-320. New York: Aca

demic Press, 1970.Courtice. F. C.. and Simmonds. W. J. Physiological significance of lymph drainage ofthe serous cavities and lungs. Physiol. Rev.. 34: 419-448. 1954.Brown. L. F., Asch, B., Harvey. V. S.. Buchinski, B.. and Dvorak. H. F Fibrinogenintlux and accumulation of cross-linked fibrin in mouse carcinomas. Cancer Res.. 48:1920-1925. 1988.

Fraker, P. J.. and Speck, J. C. Protein and cell membrane iodinations with a sparinglysoluble chloroamide, l.3.4.6-tetrachloro-3a,6a-diphenylglycoluril. Biochem. Bio-phys. Res. Commun.. SO: 849-857. 1978.

Pareira. M. D.. Sicher. N., and Lang. S. Survival of the undernourished organism. II.Tolerance of undernourished rats to rapid hemorrhage with blood volume correlations.Ann. Surg.. 147: 56-61, 1958.

Bradford. M. M. A rapid and sensitive method for the quantitation of microgramquantities of protein using the principle of protein-dye binding. Anal. Biochem., 72:248-254, 1976.

Morgan, A. G., and Terry. S. I. Impaired peritoneal fluid drainage in nephrogenicascites. Clin. Nephrol.. 15: 61-65, 1981.Dykes. P. W.. and Jones, J. H. Albumin exchange between plasma and ascitic fluid.Clin. Sci.. 34: 185-197, 1968.Counice. F. C.. and Steinbeck, A. W. The effects of lymphatic obstruction and ofposture on the absorption of protein from the peritoneal cavity. Aust. J. Exp. Biol.Med. Sci., 29:451-458. 1951.Nicoli. P. A., and Taylor. A. E. Lymph formation and flow. Annu. Rev. Physiol., 39:73-95, 1977.Shipley. R. A., and Clark. R. E.. Companmemal analysis: three-pool open systems. In:Tracer Methods for in Vim Kinetics, pp. 45-64. New York: Academic Press. 1972.

Altman. P. L. In: D. S. Dittmer (ed.). Blood and Other Body Ruids. p. 5. Washington.DC: Federation of American Societies for Experimental Biology, 1961.Lifshitz, S. Ascites, pathophysiology and control measures. Int. J. Radial. Oncol. Biol.Phys.. 8: 1423-1426. 1982.

Pust. A. H.. Leypoldt. J. K.. Frigon. R. P.. and Henderson, L. W. Peritoneal dialysatevolume determined by indicator dilution measurements. Kidney Int.. 33: 64-70. 1988.Richtmyer. R. D.. and Monon, K. W. In: Difference Methods for Initial-Value Problems, p. 142. New York: Wiley-lmerscience. 1967.Seymour. L. W.. Duncan. R., Strohalm, J.. and Kopecek. J. Effect of molecular weighl(Mw) of /V-(2-hydroxypropyl)methacrylamide copolymeni on body distribution and

rate of excretion after subcutaneous, intraperitoneal. and intravenous administration torats. J. Biomed. Mater. Res., 21: 1341-1358, 1987.Ruiili. G.. and Artbrs. K. E. Fluorescein-labelled dextran measurement in interstitialfluid in studies of macromolecular permeability. Microvasc. Res., 12: 221-230. 1976.Anurson. G.. and Wallenius. G. The renal clearance of dextran of different molecularsizes in normal humans. Scand. J. Clin. Lab. Invest.. /: 81-86, 1964.Kruger, S.. Greve. D. W., and Schueler. F. W. The absorption of fluid from theperitoneal cavity. Arch. Int. Pharmacodyn. Ther.. 137: 173-178, 1962.

Lili. S. R.. Parsons, R. H.. and Buhac. I. Permeability of the diaphragm and fluidrésorptionfrom the peritoneal cavity in the rat. Gastroenterology. 76: 997-1001,

1979.Allen. L.. and Raybuck. H. E. The effects of obliteration of the diaphragmaticlymphatic plexus on serous fluid. Anat. Ree., 137: 25-32. 1960.

Nolph, K. D.. Maclier. R.. Khanna. R.. Twardowski. Z. J., Moore, H.. and McGary.T. The kinetics of ulii.ililn.iinm during peritoneal dialysis: the role of lymphatics.Kidney Int.. 32: 219-226. 1987.Tsilibary. E. C.. and Wissig. S. L. Light and electron microscope observations of thelymphatic drainage units of the peritoneal cavity of rodents. Am. J. Anal.. 180:195-207, 1987.

Tsilibary, E. C., and Wissig, S. L. Lymphatic absorption from the peritoneal cavity:regulation of patency of mesothelial stornata. Microvasc. Res.. 25: 22-39, 1983.Ozols. R. F.. Grotzinger. K. R., Fisher. R. !.. Myers. C. E.. and Young, R. C. Kinetic-

characterization and response to chemotherapy in a transplantable murine ovariancancer. Cancer Res., 39: 3202-3208. 1979.

Abemethy, N. J., Chin. W., Hay, J. B.. Rodela. H.. Oreopoulos. D.. and Johnston, M.G. Lymphatic removal of dialysate from the peritoneal cavity of anesthetized sheep.Kidney Int.. 40: 174-181, 1991.Haraldsson. B. Diffusional transpon of albumin from inlerstitium to blood acrosssmall pores in the capillary walls of rat skeletal muscle. Acta Physiol. Scand., 133:63-71. 1988.Raybuck, H. E.. Allen, L.. and Harms. W. S. Absorption of serum from the peritonealcavity. Am. J. Physiol.. /99: 1021-1024. 1960.Barber. B. J., Schultz. T. J.. and Randlett. D. L. Comparative analysis of proteincontent in rat mesenteric tissue, peritoneal fluid, and plasma. Am. J. Physiol., 258:G714-G718. 1990.McGowan. L.. Davis, R. H.. and Bunnag. B. The biochemical diagnosis of ovariancancer. Am. J. Obstet. Gynecol.. 116: 760-768. 1973.Counice, F. C., and Robens, D. C. K. Peritoneal fluid in the rabbit: permeability ofthe mesothelium to proteins, lipoproteins and acid hydrolases. Lymphology, 8: 1-10,1975.

Research. on October 17, 2020. © 1993 American Association for Cancercancerres.aacrjournals.org Downloaded from

PATHtX3F.NF.SIS OF MALIGNANT ASCITES

70. Maurer, F. W.. Warren. M. F.. and Drinker. C. K. The composition oí"mammalian loncal dialysis. In: K. D. Nolph (éd.).Periioneal Dialysis, pp. 28—47.Amsterdam:

pericardia! and peritoneal tluid.s. Studies of their protein and chloride contents and the Martinus Nijhofi, 1989.passage of foreign substances from the blood stream into these fluids. Am. J. Physiol.. 75. Mactier, R. A., and Khanna. R. Peritoneal cavity lymphatics. In: K. D. Nolph (ed.),12V: 635-644. 1940. Peritoneal Dialysis, pp. 48-66. Amsterdam: Martinus Nijhoff, 1989.

71. Yeo, T. K., Wang, H., Nagy. J. A.. Senger, D. R.. Dvorak, H. F., and Yeo. K. T. 76. Aune. S. Transperitoneal exchange I. Peritoneal permeability studied by transperito-

Vascular permeability factor levels in line 1 and line 10 ascites tumors and human neal plasma clearance of urea. PAH. inulin. and serum albumin in rabbits. Scand. J.fluids. J. Cell Bioi., ÕI5: 421a, 1991. Gastroenterol., S: 85-97. 1970.

72. Dvorak. H. F.. Sioussal. T. M., Brown. L. F.. Berse. B., Nagy. J. A.. Sotrel. A., 77. Aune. S. Transperitoneal exchange III. The influence of transperitoneal fluid flux onManseau. E. J., Van De Water. L.. and Senger. D. R. Distribution of vascular perme- the peritoneal plasma clearance of serum albumin in rabbits. Scand. J. Gastroenterol..ability factor (vascular endothelial growth factor) in tumors: concentration in tumor 5: 161-168. 1970.blood vessels. J. Exp. Med.. 174: 1275-1278. 1991. 78. Aune. S. Transperitoneal exchange IV. The effect of Iranspcritoneal fluid transport on

73. Nagy. J. A., Masse. E. M.. Harvey-Bliss. VS.. Meyers, M.S.. Sioussal. T. M.. Senger. the transfer of solutes. Scand. J. Gastroenterol.. 5: 241-252, 1970.

D. R.. and Dvorak. H. F. Immunochemical localization of vascular permeability factor 79. Courtice, F. C. and Steinbeck, A. W. The lymphatic drainage of plasma from the(vascular endothelial growth factor) in ascites tumors: distribution in peritoneal wall peritoneal cavity of the cat. Aust. J. Exp. Biol. Med. Sci., 28: 161-169, 1950.

microvasculature. J. Cell Biol.. 115: 264a, 1991. 80. Seames. E. L.. Moncrief. J. W.. and Popovich, R. P. A distributed model of fluid and74. Korthuis. R. J.. and Granger. D. N. Role of the peritoneal microcirculation in peri- mass transfer in peritoneal dialysis. Am. J. Physiol.. 25K: R958-R972. 1990.

2643

Research. on October 17, 2020. © 1993 American Association for Cancercancerres.aacrjournals.org Downloaded from

1993;53:2631-2643. Cancer Res   Janice A. Nagy, Kemp T. Herzberg, Jane M. Dvorak, et al.   That Lead to Fluid AccumulationPathogenesis of Malignant Ascites Formation: Initiating Events

  Updated version

  http://cancerres.aacrjournals.org/content/53/11/2631

Access the most recent version of this article at:

  E-mail alerts related to this article or journal.Sign up to receive free email-alerts

  Subscriptions

Reprints and

  [email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

  Permissions

  Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/53/11/2631To request permission to re-use all or part of this article, use this link

Research. on October 17, 2020. © 1993 American Association for Cancercancerres.aacrjournals.org Downloaded from