Pathogenesis of malignant ascites: Starling's law of capillary

Annals of Oncology 12: 1353-1357. 2001.© 2001 Kluwei Academic Publishers Printed in the Netherlands.

Review

Pathogenesis of malignant ascites: Starling's law of capillaryhemodynamics revisited

J.T.Tamsma,1 H.J. Keizer2t & A. E. Meinders1Department of General Internal Medicine, 'Department of Clinical Oncology, Leiden University Medical Center, Leiden, The Netherlands,^Deceased (see Addendum)

Summary

Peritonitis carcinomatosa, indicating the presence of malignantcells in the peritoneal cavity, is a well-known complication ofmalignant disease. As a result, so-called malignant ascitesdevelops. Malignant ascites is a debilitating condition forwhich no effective anti-tumor therapy is available. Frequentdraining may be necessary to relieve pain and discomfort.Most studies regarding malignant ascites focus on diagnosisand treatment. In this paper, we will address the subject froma pathophysiologic perspective, using the characteristics ofmalignant ascites. Starling's equation of capillary forces, andrecent knowledge regarding biologically active peptides pro-duced by tumor cells. Following this approach, apart fromdecreased lymphatic ascites absorption, increased net capil-

lary fluid-production can be identified as a contributing fea-ture of ascites formation. The increased net filtration is due toan increase of overall capillary membrane-surface, increasedcapillary permeability and a subsequent increase of intra-peritoneal protein concentration leading to increased intra-peritoneal oncotic pressure. This sequence might be the resultof biologically active peptides produced by tumor cells such asvascular endothelial growth factor and basic fibroblast growthfactor. Interference with these mediators may serve as a targetin future therapeutic strategies.

Key words: angiogenesis. ascites, basic fibroblast growth factor,malignancy, metastasis, review. Starling's forces of capillaryhemodynamics, vascular endothelial growth factor

Introduction

Peritonitis carcinomatosa, indicating the presence ofmalignant cells in the peritoneal cavity, is a well-knowncomplication of malignant disease. As a result, so-calledmalignant ascites develops. Malignant ascites is a debil-itating condition for which no effective anti-tumor ther-apy is available. Frequent draining may be necessary torelieve pain and discomfort. Most studies regardingmalignant ascites focus on diagnosis and treatment. Inthis paper, we will address the subject from a pathophy-siologic perspective. First, we will review the complexmicroscopic anatomy and physiology of the normalperitoneal membrane. Secondly, characteristics of ma-lignant ascites and its pathophysiology will be reviewedusing Starling's equation of capillary forces. Finally,potential new treatment approaches will be suggested.

Anatomic and physiologic considerations

The analomv of the peritoneal membrane

A basic principle of capillary fluid hemodynamics is therelative capillary impermeabilty to proteins while fluidand solutes are able to pass the membrane relativelyeasily. As a consequence, differences in protein concen-

trations across the capillary membrane are present andoncotic pressure differences are created, necessary toreabsorb fluid from the interstitial space, thereby limit-ing net capillary fluid-filtration and preventing edemaformation. The anatomic properties of the capillary wallmeet these physiologic requirements. The microscopicanatomy of peritoneal membrane shows, apart from thecapillary endothelium and basement membrane, threedistinct barriers to prevent the loss of proteins into theperitoneal cavity: the interstitial stroma, the mesothelialbasement membrane and the mesothelial cells lining theperitoneum [1]. A schematic drawing is shown inFigure 1 [2].

Following the route from the intravascular to theintraperitoneal space (Figure 1, arrow), the endothelialcells are the first barrier encountered. Endothelial cellshave an extracellular glycocalix with fixed anioniccharges, which is difficult to pass for anionic macro-molecules such as albumin, an important contributor toplasma oncotic pressure. Peritoneal endothelial cells arelinked with tight junctions; as a result transport is trans-endothelial using intracellular pores [3, 4]. Endothelialcells are separated from the interstitial space by theendothelial basement membrane. The general plan forbasement membranes is a core of collagen to whichseveral different types of macromolecules are anchored.Proteoglycans present in the basement membrane con-

Downloaded from https://academic.oup.com/annonc/article-abstract/12/10/1353/147821by gueston 09 April 2018

1354

peritoneal cavity

endothehalcell

basementmembrane

Figure I Fluid and solute reaching the peritoneal cavity have passedthrough five anatomically distinct barriers (arrow), capillary endothe-lium and basement membrane, the interstitial stroma. the mesothelialbasement membrane and the mesothelial cells lining the peritoneum[1.2]

stitute a net negative charge, which again is a selectivebarrier for anionic proteins. The interstitial space consistsof loose connective tissue, which is composed of fibro-blasts, collagen, hyaluron-acid and negatively chargedmacromolecules. Hyualuron-acid is able to bind a con-siderable amount of water, for instance edema duringperitonitis. The interstitial space acts as a filter withsignificant resistance to diffusion of macromolecules.The submesothelial basement membrane normally ap-pears as a continuous layer at the interstitial site of themesothelial cells. Evidence is available that negativelycharged glycosaminoglycans are also present at this site.Mesothelial cells are the last barrier to be passed. Themesothelium consists of a monolayer of flat cells witha total estimated surface of approximately two squaremeters. The mesothelial cells show some functional sim-ilarity to endothelial cells in that they have a glycocalixcontaining anionic charges and transcellular channelsfor macromolecular transport [5]. In short, the presenceof thight junctions between the endothelial cells in theperitoneal capillaries, and the presence of negativelycharged macromolecules at several extracellular sitesproduce an effective barrier against leakage of nega-tively charged molecules such as albumin from plasmato the peritoneal cavity. Thus, a basic requirement toprevent excessive fluid-filtration from the capillaries tothe peritoneum is met.

The peritoneal lymphatic system

The lymphatic system collects fluid, proteins, othermacromolecules and cells, to return them to systemiccirculation. The smallest lymphatics consist of one layerof endothelium and drain into lymphatic capillaries. Abasement membrane may be present at this level but ifso, it is interrupted. The lymphatic capillary net isorganized as a plexus along the submesothelial surfaceand drains to lymphvessels. Lymphvessels have valvesand spiral-formed, smooth muscle cells and are inner-vated. Contractions of lymphvessels are generated bymyogenic stimuli, and are at least influenced by activa-tion of a-adrenoreceptors, temperature, calcium con-centrations, and vasoactive peptides. A specialised, in-triguing anatomic feature of the peritoneal lymphaticsystem are the so-called stomata. Stomata serve foropen communications between the abdominal cavityand the submesothelial diaphragmatic lymphatics. Theyare supposed to play a major role in peritoneal lym-phatic drainage [1], since most intraperitoneal fluid isabsorbed at this site [6].

The mechanisms involved in lymph formation arestill unclear. A hydraulic pressure theory has been pro-posed [7]. Normally, the interstitial pressure is negative[8] and an increase in intra-abdominal pressure will resultin increased lymph production. A close correlation be-tween fluid absorption and intra-abdominal pressurehas been shown, in line with this theory [9]. Anotherhypothesis states that osmotic forces are dominant. Thistheory postulates a protein concentrating mechanism atthe initial lymphatics [10]. Active transendothelial trans-port of albumin has been shown [11], which could createthe necessary osmotic force.

Characteristics of malignant ascites - intraperitonealprotein accumulation

Malignant ascites is characterised by positive cytologyof malignant cells. Compared to ascites caused by cir-rhosis, more white blood cells and a higher lactatedehydrogenase level are present [12, 13]. Interestingly,from the viewpoint of capillary hemodynamics, has beenthe observation that mean ascitic fluid protein-levels arehigh in patients with peritonitis carcinomatosa [13], asare ascites albumin concentrations [12-14]. Further-more, the difference between serum and ascites albuminconcentration is slight. These data show intraperitonealprotein and albumin accumulation in malignant ascitesand will be further reviewed as we discuss Starling's lawof capillary hemodynamics.

Impaired drainage or increased production?

Fluid accumulation will occur if lymphatic drainage ofthe peritoneal cavity is compromised or if net filtrationis increased, overwhelming lymphatic capacity. Perito-


1355

neal fluid kinetics has been extensively studied in peri-toneal dialysis. In dialysis, fluid accumulation is onlypossible if net filtration exceeds net absorption. Netfluid-filtration is the result of the osmolality of thedialysate. The higher the osmolality, the higher the forcethat attracts fluid from the intravascular compartment.Interestingly, osmolality of the dialysate changes in time.Osmotically active molecules disappear through lym-phatic transport and are diluted due to the attractedamount of fluid (water). An important consequence ofthis mechanism is a reduction in the rate of filtration intime [15]. In contrast, lymphatic drainage proceeds at afairly constant rate of 40 ml/h during dialysis. Thestomata, located at the peritoneal membrane lining thediaphragma, are the principle site of drainage [15]. Thus,the net effects on intraperitoneal fluid accumulation canbe calculated from the combined effects of filtration andlymphatic transport. Also, in malignant ascites, fluidaccumulation can be regarded as the result of filtrationminus drainage.

There is evidence of impaired lymphatic drainage inperitonitis carcinomatosa. This was studied in micewhen ascites was induced by injecting tumor cells intra-peritoneally [16]. Alterations in diaphragmatic, lym-phatic absorption were determined radiographically.Diaphragmatic and retrosternal lymph vessels becameoccluded five days after tumor cell injection. Ascitesformation was evident five to seven days after injectionof tumor cells. Comparable experimental data showingdecreased lymphatic drainage have been produced byothers [17]. Furthermore, lymphoscintigraphy showeddecreased lymphatic drainage in humans [18, 19]. To-gether, there is fair evidence of decreased lymphaticdrainage as a contributing factor in the pathogenesis ofmalignant ascites (Figure 2) [19].

In addition to impaired lymphatic drainage, thereis evidence of increased fluid production. Using radio-active isotopes, it was shown that the inflow rate ofplasma into the peritoneal cavity was increased six- tosixteen-fold [20]. The pathophysiology of increased fluidproduction is described by Starling's law of capillaryhemodynamics.

Pathophysiology - Starling's law of capillaryhemodynamics

The exchange of fluid between the plasma and theinterstitium is determined by the hydraulic and oncoticpressure in each compartment. The relationship betweenthese parameters can be expressed by Starling's law [21]:

Net filtration

= LpS (5 hydraulic pressure - 5 oncotic pressure).

= LpS[(P c a p-P1r)-s(7t c a p-7t l f)] .

In this equation, Lp is the unit permeability or porosityof the capillary wall, S is the surface area available for

filtration, Pcap and P,t- are the capillary and interstitialfluid hydraulic pressures, ncap and nlf are the capillaryand interstitial fluid oncotic pressures, and s representsthe reflection coefficient of proteins across the capillarywall (with values ranging from 0, if completely perme-able, to 1 if completely impermeable) [21]. Increasedcapillary permeability, increased surface area availablefor filtration, increased hydraulic pressure difference, adecreased oncotic pressure difference or a combinationof these factors could account for an increase of netfiltration.

A. Increased capillary permeability:'In peritonits carcinomatosa, increased permeability toproteins was observed in mice after intraperitoneal ad-ministration of tumor cells [16]. In another study, it wasshown that a few days after intraperitoneal injection of'Walker 256', carcinoma cells' new capillaries were ob-served and bloody ascites developed. Inhibition of angio-genesis with locally administered protamine preventednew capillaries from developing and also prevented theoccurence of bloody ascites [22]. The coincidence ofincreased permeability and new vessel formation isstriking.

Malignant cells

Production ofGrowth Factors,

eg VEGFb-FGF1

Increase of size, numberand cross-sectional area of

microvessels lining thepentoneal cavity

Accumulation in subdiaphragmottcspecialized lymphatic structures

(stomata) and lymphvossels

Figure 2 In this schematic drawing, the proposed pathogenesis ofmalignant ascites is summarized. The presence of tumor cells results inthe obliteration of lymphatic drainage as depicted in the right arm ofthe figure Furthermore, production oflocally active molecules such asVEGF and b-FGF" results in changes in ^Starling's law of capillaryhemodynamics: LpS [(Pcilp - P,r) - s(rccup - rc,r)]. Shown on the left sideare the profound changes in the balance between the capillary andinterstitial oncotic forces (itcap - 7t,r). The central pathway demonstrateschanges in peritoneal microvasculature, leading to an altered LpSproduct. As a result, net capillary fluid-filtration increases. Increasedfiltration and decreased evacuation results in ascites.

VEGF - vascular endothehal growth factor, b-FGF - basic-fibro-blastic growth factor


1356

It is now well understood that a growing tumor isdependent on angiogenesis, the formation of new bloodvessels [23], Angiogenesis starts from stimulation of theendothelium, resulting in hyperpermeability of the en-dothelial membrane and degradation of the basementmembrane and underlying stroma. The next step is themigration and proliferation of endothelial cells, and theformation of new blood vessels and capillaries [23]. Twoimportant factors in angiogenesis are basic fibroblastgrowth factor (b-FGF) and vascular endothelial growthfactor (VEGF) [24]. VEGF was discovered as a factorcreating hyperpermeability and was initially named vas-cular permeability factor (VPF) [25]. In a mouse model,it was shown that small blood vessels lining the peritonealcavity (mesentery, peritoneal wall, diaphragm) becamehyperpermeable several days after intraperitoneal tumorcell injection. The development of hyperpermeable mi-crovessels correlated with ascites VPF concentration[26]. Most tumors express VEGF [27], including ovarian[28, 29], gastric and colon carcinomas [30]. Recently, astudy was performed confirming the presence of highVEGF concentrations in malignant ascites [31, 32]. Fur-thermore, malignant ascites production, but not tumorgrowth, was completely inhibited in mice when treatedwith function-blocking VEGF antibodies. When thetreatment was stopped, all mice developed ascites withintwo weeks [33]. These positive experimental results havebeen confirmed by others using anti-VEGF antibodies[34, 35], VEGF tyrosine kinase receptor inhibitors [35,36] or exogenous soluble human VEGF receptor [37].These data suggest that increased capillary permeabilitydue to production of locally active substances such asVEGF, can be an important factor in the pathophysiol-ogy of malignant ascites (Figure 2).

B. Increased surface area available for filtration?In mice, size and number of peritoneal lining micro-vessels and subsequently cross sectional area increasedafter intraperitoneal tumor cell injection [38]. The site ofproduction of malignant ascites has also been studied inpatients using plastic rings with absorbent paper, whichwere placed on the peritoneal tumor and tumor-freesurface. The production rate of ascites of tumor-freeomentum and small bowel surface was increased. Therate of fluid production from the tumor surface was alsohigher than the fluid production of peritoneum of thecontrol subjects, but less outspoken. The authors con-cluded that "undoubtedly fluid exuded from the tumorsurface but the lion's share came from the disease-freeperitoneum" [20]. Thus, in malignant ascites, an increasedcross sectional area of microvessels lining the peritonealcavity has been shown in an experimental setting. Inaddition, it seems that in human subjects, tumor-freeperitoneal surface is able to produce the surplus of fluidin malignant ascites [20].

C. Increased hydraulic pressure difference ?The same paper [20] reported on portal pressure incontrols and in patients with ovarium cancer with or

without ascites. A minor increase in portal vein pressurewas observed in patients with ascites.

D. Decreased oncotic pressure difference?In normal physiology, albumin is known to be an effec-tive osmol that contributes to intravascular oncoticpressure, necessary to reabsorb fluid from the interstitialspace. If the oncotic pressure difference decreases, re-absorbtion decreases and interstitial fluid accumulationresults. In peritonitis carcinomatosa, protein accumulatesintraperitoneally [12]. These intraperitoneal proteins maybe partly degraded to smaller peptides and aminoacids,which could contribute to intra-abdominal oncotic pres-sure. This situation is comparable to peritoneal dialysissolutions containing a 5% aminoacid concentration,which are very effective in forcing ultrafiltration [15].As the plasma to peritoneal oncotic pressure differencedecreases, and even becomes negative, reabsorptioninto the intravascular compartment will diminish andfluid may even be 'filtrated' into the peritoneal cavity(Figure 2).

Thus, regarding Starling's law of capillary hemody-namics, we propose that the increased capillary perme-ability is essential in the pathophysiology of malignantascites. The resulting decreased or negative oncotic pres-sure difference attracts fluid into the peritoneal cavity.As mentioned, decreased lymphatic transport will sup-port intraperitoneal fluid accumulation (Figure 2).

Conclusion

The pathogenesis of malignant ascites is beginning tobe elucidated. Decreased lymphatic absorption and in-creased fluid production can be identified as contributingfactors of ascites formation. The increased net capillaryfluid-production is due to an increase of overall capillarymembrane surface, increased capillary permeability anda subsequent increase of intraperitoneal protein concen-tration, leading to increased intraperitoneal oncoticpressure. This sequence might be the result of biologi-cally active peptides, produced by tumor cells such asvascular endothelial growth factor and basic fibroblastgrowth factor. Interference with these mediators mayserve as a target in future therapeutic strategies.

Addendum

Our co-author, Dr H. J. (Jan) Keizer died shortly afterthis paper was submitted. Despite being very ill, Jancontributed significantly to its preparation. He was aleading senior registrar of oncology in our hospital. Inaddition, he intensively participated in the national andinternational field of clinical oncology.


1357

References

1. Gotloib L, Shoslak A. The functional anatomy of the peritoneumas a dialyzing membrane. InTwardowski ZJ, Nolph KD. KhannaR (eds): Contempory Issues in Nephrology, Vol 22. PeritonealDialysis: New Concepts and Applications. New York' ChurchillLivingstone 1990; 1-29.

2. Nolph KD, Miller F, Rubin J. Popovich R New directions inperitoneal dialysis concepts and applications. Kidney Int Suppl1980;Suppl 10-SI 11-6.

3. Renkin EM Some consequences of capillary permeability tomacromolecules: Starling's hypothesis reconsidered. Am J Physi-ol 1986; 250 (5 Pt 2): H706-H710.

4. Grega GJ, Adamski SW, Dobbins DE Physiological and phar-macological evidence for the regulation of permeability Fed Proc1986; 45 (2): 96-100.

5. Gotloib L, Digenis GE, Rabinovich S et al. Ultrastructure ofnormal rabbit mesentery Nephron 1983; 34 (4): 248-55.

6. Lifshitz S. Ascites. pathophysiology and control measures. Int JRadial Oncol Biol Phys 1982; 8 (8): 1423-6.

7. Allen L. Volume and pressure changes in terminal lymphatics.AmJ Physiol 1931; 123: 3.

8. Guyton AC A concept of negative interstitial pressure based onpressures in implanted perforated capsules. Circ Res 1963; 12:399.

9. Zink J, Greenway CV. Control of ascites absorption in anesthe-tized cats: Effects of intrapentoneal pressure, protein, and furo-semide diuresis. Gastroenterology 1977. 73 (5): 1119-24.

10. Casley-Smith JR. A fine ultrastructural study of variations inprotein concentration in lacteals during compression and relaxa-tion. Lymphology 1978; 12: 59-65.

11. Shasby DM, Shasby SS. Active transendothehal transport ofalbumin. Interstitium to lumen Circ Res 1985; 57 (6): 903-8.

12. Runyon BA, Hoefs JC, Morgan TR. Ascitic fluid analysis inmalignancy-related ascites. Hepatology 1988, 8 (5): 1104-9.

13. Salerno F, Restelli B, Incerti Pet al Utility of ascitic fluid analysisin patients with malignancy-related ascites. Scand J Gastroenterol1990; 25 (3): 251-6.

14. Jungst D, Xie Y, Gerbes AL Pathophysiology of elevated ascitesfluid cholesterol in malignant ascites. Increased ascites to serumrelation of proteins and lipoproteins in patients with peritonealcarcinomatosis as compared to patients with cirrhosis of the liver.J Hepatol 1992, 14(2-3): 244-8

15. Mactier RA. Kinetics of ultrafiltration with glucose and alterna-tive osmotic agents. In Twardowski ZJ. Nolph KD. Khanna R(eds): Contemporary Issues in Nephrology. Peritoneal Dialysis,New Concepts and Applications New York. Churchill Living-stone 1990; 29-52.

16. Fastaia J, Dumont AE. Pathogenesis of ascites in mice withperitoneal carcinomatosis J Natl Cancer Inst 1976; 56 (3): 547-50.

17. Feldman GB, Knapp RC, Order SE, Hellman S. The role oflymphatic obstruction in the formation of ascites in a murineovarian carcinoma. Cancer Res 1972, 32 (8): 1663-6.

18. Coates G, Bush RS. Aspin N. A study of ascites using lympho-scintigraphy with 99m Tc-sulfur colloid. Radiology 1973; 107 (3):577-83.

19. Bronskill MJ, Bush RS, Ege GN. A quantitative measurement ofperitoneal drainage in malignant ascites. Cancer 1977; 40 (5):2375-80.

20. Hirabayashi K, Graham J. Genesis of ascites in ovarian cancer.Am J Obstet Gynecol 1970; 106 (4) 492-7

21. Rose BD, Post TW. Edematous states. In Rose BD. Post TW(eds): Clinical Physiology of Acid-base and Electrolyte Disorders.New York: McGraw-Hill 2001; 478-534.

22. Heuser LS, Taylor SH. Folkman J Prevention of carcinomatosisand bloody malignant ascites in the rat by an inhibitor of angio-genesis. J Surg Res 1984; 36 (3): 244-50.

23. Hanahan D. Folkman J. Patterns and emerging mechanisms ofthe angiogenic switch during lumongenesis. Cell 1996; 86 (3):353-64.

24. Neufeld G. Cohen T. Gengnnovitch S. Poltorak Z. Vascularendothehal growth factor (VEGF) and its receptors. FASEB J1999; 13(1): 9-22.

25. Senger DR. Galli SJ. Dvorak AM et al. Tumor cells secrete avascular permeability factor that promotes accumulation of ascitesfluid. Science 1983: 219 (4587): 983-5.

26 Nagy JA. Masse EM. Herzberg KTet al. Pathogenesis of ascitestumor growth: Vascular permeability factor, vascular hyper-permeability, and ascites fluid accumulation. Cancer Res 1995: 55(2): 360-8.

27. Senger DR. Perruzzi CA. Feder J. Dvorak HF. A highly conservedvascular permeability factor secreted by a variety of human androdent tumor cell lines. Cancer Res 1986:46(11) 5629-32.

28. Yamamoto S. Konishi I, Mandai M et al. Expression of vascularendothelial growth factor (VEGF) in epithelial ovarian neo-plasms: Correlation with clinicopathology and patient survival,and analysis of serum VEGF levels. Br J Cancer 1997: 76 (9):1221-7.

29. Barton DP. Cai A. Wendt K. Young M et al Angiogenic proteinexpression in advanced epithelial ovarian cancer. Clin CancerRes 1997: 3 (9): 1579-86.

30. Zebrowski BK. Liu W. Ramirez K et al. Markedly elevated levelsof vascular endothelial growth factor in malignant ascites. AnnSurg Oncol 1999; 6 (4) 373-8.

31. Kraft A. Weindel K. Ochs A et al Vascular endothelial growthfactor in the sera and effusions of patients with malignant andnonmalignant disease. Cancer 1999; 85 (I): 178-87

32. Santin AD. Hermonat PL. Ravaggi A et al. Secretion of vascularendothelial growth factor in ovarian cancer Eur J GynaccolOncol 1999; 20 (3). 177-81.

33. Mesiano S, Ferrara N. Jafl'e RB. Role of vascular endothelialgrowth factor in ovarian cancer: Inhibition of ascites formationby immune-neutralization. Am J Pathol 1998; 153 (4). 1249-56.

34. Luo JC. Toyoda M. Shibuya M. Differential inhibition of fluidaccumulation and tumor growth in Iwo mouse ascites tumors byan antivascular endothelial growth factor/permeability factorneutralizing antibody Cancer Res 1998. 58 (12): 2594-600.

35. Schlaeppi JM. Wood JM. Targeting vascular endothelial growthfactor (VEGF) for anti-tumor therapy, by anti-VEGF neutraliz-ing monoclonal antibodies or by VEGF receptor tyrosinc-kinaseinhibitors. Cancer Metastasis Rev 1999: 18 (4): 473-81

36 Xu L. Yoneda J. Herrera C et al. Inhibition of malignant ascitesand growth of human ovarian carcinoma by oral administrationof a potent inhibitor of the vascular endothelial growth factorreceptor tyrosine kinases. Int J Oncol 2000: 16 (3): 445-54.

37. Stoelcker B. Echtenacher B. Weich HA et al. VEGF/Flk-I inter-action, a requirement for malignant ascites recurrence. J Inter-feron Cytokine Res 2000: 20 (5). 511-7

38. Nagy JA. Morgan ES. Herzberg KTct al. Pathogenesis of ascitestumor growth: Angiogenesis. vascular remodeling, and stromaformation in the peritoneal lining. Cancer Res 1995: 55 (2)- 376-85.

Received 11 April 2001; accepted 22 June 2001

Correspondence toJ T Tamsma. MDLeiden University Medical Center (LUMC)Dept. of General Internal Medicine. Cl-RPO Box 96002300 RC LeidenThe NetherlandsE-mail: jttamsma(<ilumc.nl


Documents

Pathogenesis of malignant ascites: Starling's law of capillary