Effects of the taxanes paclitaxel and docetaxel on edema formation and interstitial fluid pressure

Aurora Brønstad, Ansgar Berg, Rolf K. Reed


Interstitial fluid pressure (Pif) is important for maintaining constant interstitial fluid volume. In several acute inflammatory reactions, a dramatic lowering of Pif has been observed, increasing transcapillary filtration pressure and favoring initial and rapid edema formation. This lowering of Pif seems to involve dynamic β1-integrin-mediated interactions between connective tissue cells and extracellular matrix (ECM) fibers. β1-Integrins are adhesion receptors responsible for the attachment of connective tissue cells to the ECM providing a force-transmitting physical link between the ECM and cytoskeleton. Disruption of actin filaments leads to lowering of Pif and edema formation, suggesting a role for actin filaments. The aim of this study was to further investigate the role of the cytoskeleton in the control of Pif by studying the effect of microtubuli fixation using paclitaxel and docetaxel. Pif was measured with the micropuncture technique. Albumin extravasation (Ealb) was measured using 125I-labeled albumin. Paclitaxel and docetaxel were tested locally on foot skin in female Wistar rats. Paclitaxel (6 mg/ml) reduced Pif from −1.5 ± 1.0 mmHg in controls to −4.9 ± 2.6 mmHg after 30 min (P < 0.05) in a dose-dependent manner (P < 0.05). Docetaxel caused a similar lowering of Pif. Both paclitaxel and docetaxel increased Ealb compared with Cremophor EL and saline control (P < 0.05). Pretreatment with phalloidin before paclitaxel, causing fixation of actin filaments, abolished the lowering of Pif caused by paclitaxel. This study confirms several previous studies demonstrating that connective tissue cells influence Pif and edema formation.

  • capillary permeability
  • inflammation
  • connective tissue
  • microtubuli

interstitial fluid pressure (Pif) is an important factor in the control of interstitial fluid volume (IFV) (2). Lowering of Pif facilitates fluid flux across the capillary wall and thereby contributes significantly to initial and rapid edema formation during acute inflammation (28). There is an increasing amount of evidence demonstrating that connective tissue cells play a role in the control of Pif and IFV through interactions via cell surface receptors toward extracellular matrix (ECM) components. The adhesion receptors toward the ECM are called β1-integrins. Blockade of integrin function in vivo with antibodies toward β1-integrins resulted in lowering of Pif and edema formation (29, 30). β1-Integrin function can also be modulated in vivo through intracellular pathways (11).

Actin filaments are structural cell cytoskeletal proteins important for the contractile apparatus of the cell and for integrin function (5). Earlier experiments have shown that cytochalasin D, which blocks formation of actin filaments, lowers Pif and causes edema when injected subcutaneously (3). Furthermore, phalloidin, which causes fixation of actin filaments by preventing their depolymerization, abolished the lowering of Pif and attenuated edema formation induced by dextran anaphylaxis in the Wistar rat (4). The microtubuli are a second type of cytoskeletal filaments, and they have been implicated in a variety of cellular phenomena such as intracellular vesicle transport and cilia motility. Microtubuli are built up by polymerization of tubulin monomers, and they are labile structures that are continuously reorganized (1, 12). Subcutaneous injection of nocodazole and colchicine, which prevent polymerization of tubulin monomers by binding to tubulin, had no effect on Pif in similar experiments (3). As a continuation of earlier experiments from our group, we wanted to investigate the effect of microtubuli fixation using a group of substances called “taxanes.” Taxanes bind tightly and rapidly to microtubuli and thereby prevent disassembly of filaments (1). Taxol (paclitaxel) and Taxotere (docetaxel) are commercially available antimitotic drugs used in the treatment of cancer in humans. Docetaxel is more potent than paclitaxel in inducing microtubuli bundling (22). Edema and anaphylactic reactions are reported adverse effects of these drugs (25). Cremophor EL, the vehicle and solvent for paclitaxel in Taxol, has also been claimed to cause the inflammatory edema and histamine release and is suggested to be responsible for at least some of the adverse effects of this agent (20).

The aim of this study was to explore further the role of the cytoskeleton in the control of Pif by hypothesizing that the effect of paclitaxel and docetaxel would be to lower Pif and induce albumin extravasation (Ealb) and edema formation.



Female Wistar rats, weighing 200–250 g (M&B; Ejby, Denmark), were used in the experiments. They were housed five per cage (Macrolon IV) and fed a standard diet and tap water ad libitum for at least 10 days after arrival at the animal facility. The animals were randomized before they took part in the experiment. The rats were anesthetized with an intraperitoneal injection of pentobarbital (70 mg/kg body wt) before surgery and throughout the experiment. They were kept in dorsal recumbence on a heating pad to maintain body temperature at 37.5–38.5°C. The experimental protocol and procedures were approved by and performed in accordance with the regulations laid down by the National Animal Research Authority.

Test Substances

Paclitaxel (Taxol, Bristol-Meyers-Squibb; München, Germany) was used undiluted as a 6 mg/ml stock solution or diluted in sterile saline (0.9% NaCl) to 0.6 and 0.06 mg/ml. Docetaxel (Taxotere, 40 mg/ml, Aventis Pharma; London, UK) was prepared according to the manufacturer's instructions, diluted to 0.5 mg/ml, and used within 8 h after dilution. Ethanol (369 mg) was added to Cremophor EL (527 mg, Sigma; St. Louis, MO) and used as a control solution for paclitaxel. A 10-μl Exmire microsyringe (style G-30, point B needles) was used for subcutaneous administration of the test substance. Human serum albumin (HSA) labeled with 125I and 131I (125I-HSA and 131I-HSA, respectively) was supplied by Norwegian Energy Technique (Kjeller, Norway). Pentobarbital was supplied as a 50 mg/ml nonsterile solution.


Series 1: edema formation and Ealb.

A polyethylene-50 catheter was placed in the right jugular vein and used for the administration of radioactive tracers and saturated potassium chloride.


After surgical preparation (see above), 5 μl of the test substance were administered subcutaneously on the dorsal side of the foot. Thereafter, 125I-HSA was given intravenously and circulated for 25 min before the second tracer, 131I-HSA, was given. Five minutes thereafter, blood samples were taken by cardiac puncture, and circulatory arrest was induced by an intracardiac injection of 0.5 ml saturated potassium chloride. The rationale for giving two tracers is that 125I-HSA is allowed time to distribute in the extracellular and vascular space while 131I-HSA is distributed only in the vascular space. Thus the difference between their distribution space is the plasma equivalent volume extravasated during 25 min. Ten microliters of serum were diluted in 2 ml of 0.9% NaCl for the determination of radioactivity. Skin samples (dermis and subcutis) were taken from the dorsal aspect of both feet using a forceps and a pair of scissors. The tissue samples were put directly into preweighed vials and corked. Radioactivity was measured with a gamma-counter (Wallac Wizard, 1470 automatic gamma counter, Perkin Elmer) with automatic spillover and background correction. Distribution volumes were calculated as plasma-equivalent space (counts per min per gram dry tissue weight divided by counts per min per ml plasma). Ealb was estimated as the 25-min extravascular distribution volume of 125I-HSA.

Total tissue water.

The skin samples were dried at 65°C and weighed twice a week until stable weight was obtained, usually after 2–3 wk. Total tissue water (TTW) in skin was calculated as water content per gram of dry tissue [(wet weight − dry weight)/(dry weight)]. There was no correction for blood volume. TTW was obtained from the same skin samples used for measurement of Ealb.

Series 2: Pif.

Circulatory arrest was induced in deep anesthesia with an intracardiac injection of 0.5 ml saturated potassium chloride and before the administration of the test substance on the foot.

Pif was measured using sharpened glass micropipettes made of glass capillaries (GC 100–15, 1.00 mm outer diameter × 0.58 mm inner diameter, Harvard Apparatus), which were pulled on a micropipette puller (P-87, Sutter Instruments) and sharpened (MB3/T-PSU5 Microbeveller, World Precision Instruments) to a final tip diameter of 4–9 μm. The pipettes were filled with 0.5 M NaCl solution colored with Evans blue solution, and the pressure was measured with a servo-controlled counterpressure unit (35, 37). Micropuncture was performed under visual guidance using a stereomicroscope (Wild M5) and a micromanipulator (Leitz). Measurements were performed through intact skin on the dorsal aspect of the left foot.

After the hair was carefully cut using a pair of scissors, the foot was gently fixed to the table with adhesive tape. Care was taken not to compress, stretch, or pinch the skin. The test substance was administered in the center of a circle that was 5 mm in diameter. The circle and its center were marked with a water-soluble ink pen after circulatory arrest, and measurements were taken during the subsequent 60 or 90 min. Measurements were taken proximally to the injection site in the periphery of the circle. The zero level was defined in a bath of isotonic saline with the surface at the same height as the point of measurement. Criteria for acceptance of the measurement have been described elsewhere (4, 37). Control Pif was measured before circulatory arrest.

Experimental Protocol

Series 1: Ealb and TTW.

The animals received test substances on the hind feet according to Table 1.

View this table:
Table 1.

Experimental protocol for series 1: albumin extravasation and total tissue water (see materials and methods for further explanation)

Series 2: Pif.

All animals received 2 μl of the test substance or vehicle subcutaneously on the dorsal side of the left foot after circulatory arrest with the exception of phalloidin, which was given intravenously.


Paclitaxel was given undiluted (6 mg/ml, n = 8) at 10% (0.6 mg/ml, n = 8) and at 1% (0.06 mg/ml, n = 8). Cremophor EL served as the control for paclitaxel (n = 12). Pif was measured for the subsequent 90 min after the injection.


Docetaxel (0.5 mg/ml, n = 8) or saline (9 mg/ml, n = 8) was administered after circulatory arrest. Pif was measured for 60 min after the injection.


Phalloidin (125 μg, n = 6) in 0.25 ml saline was given intravenously and circulated 30 min before circulatory arrest was induced. Saline (2 μl, 9 mg/ml) was given subcutaneously on the dorsal aspect of the left foot. Measurements were taken for the next 60 min.

Subcutaneous docetaxel (n = 7; control group for phalloidin experiments).

Saline (0.25 ml) was given intravenously and circulated 30 min before circulatory arrest and control measurement, whereafter docetaxel (2 μl, 0.5 mg/ml) was given subcutaneously and Pif was measured for 60 min.

Docetaxel (n = 9) after pretreatment with phalloidin.

Phalloidin (125 μg) in saline (0.25 ml) was given intravenously and circulated for 30 min before control measurement and subsequent circulatory arrest was induced. Docetaxel (2 μl, 0.5 mg/ml) was given subcutaneously after circulatory arrest and measurements continued for 60 min. The control group for these experiments (n = 7) was given saline (0.25 ml) intravenously instead of phalloidin.

Statistical Methods

Results are presented as means ± SD unless otherwise stated. The changes in values were calculated as the differences between the right and left foot and are presented as means ± SD. Statistical analyses were performed with one-way ANOVA and subsequent Bonferroni or Dunn's pair-wise multiple comparisons. When data did not exhibit normal distribution, nonparametric tests were used. P < 0.05 was considered statistically significant.


Series 1: Ealb and TTW

Ealb in the foot injected with 6 mg/ml paclitaxel was 0.27 ± 0.17 ml/g dry wt, significantly higher than 0.15 ± 0.10 ml/g dry wt in the control group treated with Cremophor EL (P < 0.05; Table 2). Ealb and TTW in the feet receiving Cremophor EL and NaCl were not significantly different (P > 0.05) but significantly higher than the value for no injection. The difference between no injection and saline control was due to the effect of the injection trauma. Docetaxel (0.5 mg/ml) increased Ealb from 0.17 ± 0.10 ml/g dry wt in saline control to 0.44 ± 0.27 ml/g dry wt after docetaxel (P < 0.05). The corresponding TTW increased from 2.13 ± 0.20 to 2.29 ± 0.28 ml/g dry wt (P < 0.05; Table 2). When no injection was given, Ealb and TTW were 0.02 ± 0.01 and 1.92 ± 0.16 ml/g dry wt, respectively (Table 2). The increase in Ealb above the no injection group (ΔEalb) was ∼0.10 ml/g dry wt for saline and Cremophor EL, whereas for docetaxel and paclitaxel the increase averaged 0.18 and 0.10 ml/g dry wt above their respective controls (P > 0.05; Table 2).

View this table:
Table 2.

Effects of paclitaxel, cremophor EL, docetaxel, saline, and no injection on Ealb and TTW

Series 2: Pif

Lowering of Pif was not seen after the injection of Cremophor EL or saline (P > 0.05 compared with own controls). Injection of paclitaxel at 6 and 0.6 mg/ml lowered Pif significantly from about −1.5 mmHg in control to −4.9 ± 2.5 and −3.7 ± 1.6 mmHg at 60 min, respectively (P < 0.05 for both time intervals compared with own control; Fig. 1). Paclitaxel (0.06 mg/ml) caused significant lowering of Pif to −3.9 ± 1.8 mmHg after 60 min compared with −1.4 ± 1.8 mmHg with Cremophor EL (P < 0.05). The lowering of Pif was dose dependent (Fig. 1) and ranged from −2.5 mmHg down to −11 mmHg in individual experiments. Pif in the foot receiving docetaxel fell from −0.8 ± 1.6 mmHg in control to −3.2 ± 2.4 mmHg within the first 30 min postinjection (P < 0.05) and further to −4.0 ± 2.4 mmHg after the following 30 min (P < 0.05; Fig. 2). Finally, pretreatment with phalloidin totally abolished the lowering of Pif induced by docetaxel (P < 0.05; Fig. 3).

Fig. 1.

Interstitial fluid pressure (Pif) after paclitaxel concentrate (cons), 10% paclitaxel, 1% paclitaxel, and Cremophor EL. Values are means ± SE. *P < 0.05 compared with control; #P < 0.05 compared with Cremophor EL.

Fig. 2.

Pif after docetaxel and physiological saline. Values are means ± SE. #P < 0.05 compared with saline.

Fig. 3.

Pif after docetaxel with and without pretreatment with phalloidin. The docetaxel experiments are different from those in Fig. 2. Values are means ± SE. #P < 0.05 comparing the pretreated and untreated group.


Paclitaxel and docetaxel bind tightly to and cause “fixation” of the microtubuli. Both paclitaxel and docetaxel lowered Pif and increased Ealb. Furthermore, pretreatment with phalloidin, causing fixation of actin filaments, abolished the lowering of Pif otherwise seen after docetaxel. Docetaxel was the most potent drug in inducing capillary leakage of albumin and rapid lowering of Pif. The test substances used likely affect all cells in the tissue including the vascular cells. The effect on Pif is likely a “pure” effect of interstitial/connective tissue cells, whereas the effect on Ealb and TTW is an effect also including the vascular cells.

Increased Ealb after docetaxel has previously been demonstrated with use of a microdialysis technique (18). The same study (18) also showed that pretreatment with phalloidin abolished the increase in albumin extravasation by docetaxel.

Visible edema is not expected before a doubling of IFV has occurred (36). The corresponding increase in TTW and Ealb (Table 2) suggests that capillary permeability is increased because maintained capillary permeability with increased transcapillary fluid flux normally will yield a smaller rise in Ealb than in TTW due to the capillary sieving. Thus the corresponding increase in absolute mean values for TTW and Ealb demonstrates that the selectivity at the capillary membrane was reduced, i.e., the capillary permeability was increased (34). The average increase in TTW after paclitaxel compared with Cremophor EL averaged ∼0.14 ml/g dry wt for Ealb and 0.12 ml/g dry wt. The corresponding figures for docetaxel compared with saline were 0.16 and 0.27 ml/g dry wt, respectively, also suggesting that the capillary membrane became more leaky to albumin. For paclitaxel, the increase above that of the injection was smaller than that for docetaxel, but again, and along the same argument as above, the test substance caused a reduction in the selectivity of the capillary membrane. It should be noted that there was a significant increase in Ealb but not TTW due to injection trauma itself (Table 2). Thus the physiological effect of the substances tested should be compared with injection of respective control substances and not to “no injection.”

Cremophor EL, a derivative of castor oil and ethylene oxide, is used together with ethanol as a solvent for paclitaxel. Proinflammatory properties of Cremophor EL, including histamine release (9, 10) and complement activation (20), have been reported (7–9, 19, 20). In fact, Cremophor EL has been suggested to cause some of the hypersensitivity reactions and adverse effects in patients treated with paclitaxel (26, 33). The similarity between subcutaneous injections of saline and Cremophor EL on Pif, Ealb, and TTW clearly demonstrates that at least the parameters measured in this study were not influenced by Cremophor EL because they were not affected beyond that of injection of saline.

Semb et al. (32) reported that no change in Pif occurred in skin as measured with the wick-in-needle method after systemic paclitaxel treatment in a clinical study on cancer patients. The present study demonstrates the effect on Pif and edema formation as well as transcapillary fluid flux in the first 30–60 min after local injection of the test substance. When Pif is measured in acute inflammation, it is important to induce circulatory arrest to prevent edema formation, which potentially causes an underestimation of lowering of Pif. In the study of Semb et al. (32), Pif was in a dynamic balance between transcapillary fluid flux and lymphatic drainage of fluid. The results should therefore be different from those reported here because the lowering of Pif enhances transcapillary fluid flux, contributing to the edema formation (21). In the study of Semb et al. (32), a constant interstitial colloid osmotic pressure despite a lowering of plasma colloid osmotic pressure also suggests an increased capillary permeability (lowered selectivity of the capillary membrane). This is in agreement with the present observations of a similar absolute increase in the absolute values of Ealb and TTW (see above).

We have previously demonstrated that lowering of Pif is a phenomenon that accompanies several inflammatory reactions (28). The lowering of Pif greatly facilitates the increase in transcapillary fluid flux required to explain the rapidly appearing edema. The phenomenon seems to involve perturbation of normal β1-integrin function as a final and common step (28) either by reducing the number of attachments between the connective tissue cells and the ECM or by reducing the tension by which each cell pulls on the matrix (see below). The β1-integrin receptors are a class of heterodimeric receptors by which the cells attach to the ECM (5). The lowering of Pif is thought to be related to the ability of loose connective tissue to swell when allowed access to saline due to its content of hyaluronan and glycosaminoglycans (23). Meyer et al. (24) demonstrated that this hyaluronan/glycosaminoglycan gel is restrained by a microfilament/collagen network. Rubin et al. (31) described a model for IFV control where connective tissue cells actively compress the tissues by placing strain on the collagen and/or extracellular microfilament networks. The swelling hyaluronan is compacted, thereby preventing excess fluid from entering the interstitial fluid space (31). When the connection between the cell and ECM is perturbed, the tension exerted by the connective tissue cells is released, and the tissue can imbibe fluid due to its content of hyaluronan and glycosaminoglycans in the ECM (31). The released strain from the connective tissue cells allows the tissue to swell and fluid to enter the interstitial fluid space and form edema. The importance and role of β-integrins was documented by disruption of the cell-ECM interaction by polyclonal β-integrin antibodies and monoclonal α2β1-integrin antibodies, which resulted in lowering of Pif and edema (29).

β1-Integrin receptors are in turn attached to the actin filament part of the cytoskeleton via a series of intracellular proteins. Thus integrins mediate tension from the cell interior to the ECM. Cytoskeletal disruption using cytochalasin D, which depolymerizes actin filaments into monomers, resulted in lowering of Pif (3). Furthermore, phalloidin, which prevents depolymerization and cause fixation of actin filaments, abolished the lowering of Pif and attenuated edema formation in dextran anaphylaxis in Wistar rats, but phalloidin alone had no effect on Pif in control (4). Furthermore, the effect of fixing the filamentous actin with phalloidin versus fixing the microtubuli with taxanes actually had opposite effects on Pif and edema formation, which might seem contradictory. Also, nocodazole and colchicine, which disrupts microtubuli, were without effect on Pif (3).

Actin filaments and microtubuli as well as the intermediate filaments, which constitute the third component of the cytoskeleton, are all interconnected and act together when it comes to polarizing the cell (1). They have different functions in the cell. Actin filaments are components of the contractile apparatus in the cells, whereas microtubuli constitute the intracellular framework. Danowski (6) showed that microtubuli inhibitors stimulated fibroblast contractility. Also, microtubuli disruption by using nocodazole caused increased contractile force in pig aortic rings, whereas microtubuli fixation with taxol reduced the contractile force in aortic rings (27).

A commonly used model describing the interactions between the different components of the cytoskeleton, actin filaments, microtubuli, and intermediate filaments, was developed by Ingber and associates (13) using a tensegrity theory. This theory tried to explain and understand the relationship between mechanical stimulus and biochemical response in the cell (16). In this model, the actin filaments provide the active and contractile forces that are mediated via the integrins to the ECM (14). Furthermore, the microtubuli are stress-bearing struts and the intermediate filaments interconnect at several places between the microtubuli and actin filaments. With disruption of actin filaments with cytochalasin D, a decreased traction was observed by cells on their attachment (13, 17, 27). A lowering of Pif (3) is explained by lessened strain from the cells on the extracellular matrix, thus allowing the matrix to expand, in turn causing lowering of Pif when no fluid is added to the matrix. Along the same reasoning as above, fixation of the contractile actin filaments should not result in altered traction on the substrate on which cells are grown or on Pif. Regarding the changes subsequent to disruption of the microtubuli network, with colchicine and nocodazole, an increased traction was observed from the cells on their substrate (6, 15). This was in agreement with the model prediction that disruption of the stress-bearing struts allows more efficient contractions of actin filaments. With regard to Pif, no change was observed after nocodazole and colchicine (3), seemingly in contrast to the tensegrity model because an increase should have been expected. Nevertheless, this is in agreement with the observation that, although platelet-derived growth factor BB was able to raise Pif after it has been lowered in inflammatory reaction (30), it was not able to raise Pif from a normal and unperturbed state, possibly because an increased Pif will remove fluid from the interstitium via still-active lymphatics. Furthermore, and along the reasoning above, fixation of the microtubuli with taxanes will “increase” the capacity of the stress-bearing microtubuli/“struts.” This will cause the contractile actin filaments to become less efficient and in turn cause lowered force/stress transmitted to ECM networks and thereby allowing the matrix to swell in turn resulting in a lowering of Pif. Finally, the experiments with phalloidin pretreatment before docetaxel are also well explained in this model. When the active and contractile components are “fixed” with phalloidin, the subsequent increase in the amount/level of the stress-bearing component will not affect the cell tension that is exerted on the ECM. To summarize, the observations of Pif in the present study with taxanes and other agents that affect the cytoskeleton are in agreement with the tensegrity concept Ingber et al. (13) proposed for force transfer between the cell cytoskeleton and ECM.

The experiments with phalloidin demonstrated that when the actin filaments have been “fixed” and stabilized by phalloidin, there was no longer an effect of paclitaxel, i.e., the microtubuli network was without an effect on Pif. Reversal of the administration of phalloidin and paclitaxel was not performed because a lowering of Pif had already taken place and edema had developed. Thus microtubuli can participate in the control and modulation of Pif and thereby in transcapillary fluid and albumin transport and control of IFV.


This study received financial support from The Norwegian Research Council and The Norwegian Heart Association.


Cremophor EL was a gift from Bristol-Myers-Squibb.


  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract