Increased endothelial uptake of paclitaxel as a potential mechanism for its antiangiogenic effects: Potentiation by Cox-2 inhibition

Authors

  • Jaime R. Merchan,

    1. Division of Hematology-Oncology, Department of Medicine and the Center for Study of the Tumor Microenvironment, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA, USA
    2. The Clinical Investigator Training Program: Beth Israel Deaconess Medical Center-Harvard/MIT Health Sciences and Technology, in collaboration with Pfizer, Inc., Boston, MA, USA
    Current affiliation:
    1. Department of Medical Oncology, Mayo Clinic, 200 First Street, SW, Rochester, MN 55905
    Search for more papers by this author
  • Deepa R. Jayaram,

    1. Division of Nephrology, Department of Medicine and the Center for Study of the Tumor Microenvironment, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA, USA
    Search for more papers by this author
  • Jeffrey G. Supko,

    1. Massachusetts General Hospital and Harvard Medical School, Boston, MA, USA
    Search for more papers by this author
  • Xiaoying He,

    1. Massachusetts General Hospital and Harvard Medical School, Boston, MA, USA
    Search for more papers by this author
  • Glenn J. Bubley,

    1. Division of Hematology-Oncology, Department of Medicine and the Center for Study of the Tumor Microenvironment, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA, USA
    Search for more papers by this author
  • Vikas P. Sukhatme

    Corresponding author
    1. Division of Hematology-Oncology, Department of Medicine and the Center for Study of the Tumor Microenvironment, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA, USA
    2. Division of Nephrology, Department of Medicine and the Center for Study of the Tumor Microenvironment, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA, USA
    • Beth Israel Deaconess Medical Center, 330 Brookline Avenue, Dana 517, Boston, MA 02215
    Search for more papers by this author

Abstract

Paclitaxel has antiangiogenic properties, but the mechanisms for the enhanced sensitivity of endothelial cells (ECs) to this drug are not established. The aims of our study were to compare the distribution of paclitaxel into ECs with other cell types, to assess the effects of low doses of paclitaxel on Cox-2 expression and to determine the combined effects of paclitaxel and Cox-2 inhibitors on angiogenesis in vitro and in patients with cancer. Upon exposure to low (5 nM) concentrations of [3H]-paclitaxel, uptake of radioactivity was more than 5 times higher in ECs than other cell types. Exposing human umbilical vein ECs to low nanomolar (1–50 nM) concentrations of paclitaxel enhanced Cox-2 expression more than 2-fold, as measured by ELISA. Combined treatment with paclitaxel and the Cox-2 inhibitor NS-398 resulted in increased antiendothelial effects as compared to each agent alone. To assess the biologic effects of the combined treatment in vivo, 4 cancer patients were treated with a prolonged intravenous infusion of paclitaxel (10 mg/m2/day) and the Cox-2 inhibitor celecoxib (400 mg p.o. BID), and plasma angiogenic activity and drug levels were measured. The treatment was well tolerated, providing steady-state concentrations of paclitaxel in plasma near 10 nM and potent plasma antiendothelial effects were observed. These findings suggest that antiangiogenic effects of paclitaxel may be due its preferential accumulation in ECs. Low dose paclitaxel in combination with a Cox-2 inhibitor is an attractive antiangiogenic and antitumor strategy that deserves further evaluation in clinical trials.

Angiogenesis is a key feature for tumor growth and progression.1, 2 Many chemotherapeutic agents have been identified that possess antiangiogenic properties3, 4, 5, 6 and some are being formally evaluated in clinical trials. Among these compounds are the microtubule-interfering agents vinblastine, paclitaxel and docetaxel.3, 7, 8, 9, 10, 11, 12, 13, 14 Paclitaxel is a microtubule-polymerizing agent that is widely used in the treatment of solid tumors. In addition to its known cytotoxic effects, it inhibits endothelial cell migration, invasion, capillary tube formation and proliferation in vitro at concentrations ranging from picomolar to nanomolar, depending on the endothelial cell type and the assay conditions.3, 15, 16 In addition, paclitaxel has been shown to significantly reduce microvessel density and decrease vascular endothelial growth factor production in vivo.3, 10, 12 The mechanism(s) responsible for the exquisite sensitivity of ECs to paclitaxel have not been elucidated. Because ECs appear to be particularly sensitive to paclitaxel, we hypothesized that they may have the ability to accumulate paclitaxel to a greater degree than other cell types, which could contribute to their enhanced sensitivity to this agent.

A growing body of evidence indicates that Cox-2 plays a significant role in tumor angiogenesis and that Cox-2 inhibition suppresses tumor growth and neovascularization in different tumor types.17, 18, 19 The contributions of Cox-2 in tumor angiogenesis include increased expression of VEGF by tumor cells; the production of the eicosanoid products thromboxane A2, PGE2 and PGI2 that can directly stimulate EC migration and growth factor induced angiogenesis and potentially inhibition of EC apoptosis by stimulation of Bcl-2 or Akt activation.19, 20 Paclitaxel has been shown to induce Cox-2 mRNA expression, increase Cox-2 protein levels and enhance PGE2 production in epithelial and tumor cell lines.21 Cox-2 overproduction induced by paclitaxel may therefore cause the undesirable effect of indirectly stimulating angiogenesis by this mechanism. This led us to the hypothesis that the antiangiogenic effects of paclitaxel might be enhanced by Cox-2 inhibition.

This report shows that the enhanced sensitivity of ECs to low concentrations of paclitaxel may be due to their increased ability to accumulate the drug over non-ECs. It is also demonstrated that low concentrations of paclitaxel increases Cox-2 expression in ECs and simultaneous treatment with a Cox-2 inhibitor enhances the antiproliferative activity of paclitaxel against stimulated ECs. These findings were extended to the clinical setting, where patients with prostate cancer or malignant melanoma were treated with low dose paclitaxel given as a prolonged continuous i.v. infusion together with the specific Cox-2 inhibitor celecoxib given orally. Antiangiogenic activity was detected in the plasma specimens obtained from the patients during therapy, as measured by in vitro bioassays. These findings suggest that treatment with low dose paclitaxel in combination with a Cox-2 inhibitor may be an attractive clinical strategy to enhance the intrinsic antiangiogenic and antitumor effects of each agent alone.

Abbreviations:

bFGF, basic fibroblast growth factor; Cox-2, cyclooxygenase-2; EC, endothelial cell; HMVEC-D, human microvascular endothelial cells-Dermal; HMVEC-L, human microvascular endothelial cells-Lung; HUVEC, human umbilical vein endothelial cells; PSA, prostate specific antigen; VEGF, vascular endothelial growth factor.

Material and methods

Reagents

The clinical formulation of paclitaxel (Taxol®, Bristol Myers Squibb, Princeton, NJ) a 6 mg/ml (7.02 mM) solution in Cremophor EL and dehydrated alcohol, was diluted in sterile phosphate-buffered saline (PBS) to the desired concentrations and used for the bioassays. [o-benzamido-3H]-paclitaxel (specific activity, 37 Ci/mmol) was purchased from Moravek Biochemicals (Brea, CA). The Cox-2 inhibitor NS-398 (Sigma-Aldrich Research, St. Louis, MO) was dissolved in sterile DMSO (Sigma-Aldrich Research, St. Louis, MO) to a concentration of 1 mM and then diluted in sterile PBS to the desired working concentrations. The clinical formulation of celecoxib (Celebrex, Pfizer, Groton, CT) was used for the clinical trials (pure celecoxib was not available at the time of the in vitro studies). Matrigel (Collaborative Biomedical Products, Bedford, MA) was used at 7 mg/ml for in vitro angiogenesis (tube formation) assays. Basic fibroblast growth factor and vascular endothelial growth factor were purchased from R&D Systems (Minneapolis, MN). The cell proliferation reagent WST-1 (Roche, Indianapolis, IN) was used for proliferation assays. Additional solvents, reagents and chemicals were obtained from commercial sources in grades appropriate for direct use unless otherwise specified.

Cell culture

Human umbilical vein endothelial (HUVE) cells, human lung microvascular endothelial cells (HMVEC-L) and human dermal microvascular endothelial cells (HMVEC-D) were obtained from Clontech Lab, Inc. (Palo Alto, Ca) and used between passages 3 and 5. They were maintained in EGM2-MV medium (Clontech, Palo Alto, Ca). Primary human fibroblasts (IMR-90) were obtained from ATCC (Manassas, VA) and maintained in DMEM supplemented with 10% fetal bovine serum (FBS) and antibiotics. The human renal carcinoma (786-O) and prostate cancer (PC-3) cell lines were obtained from ATCC. PC-3 cells were maintained in RPMI medium supplemented with 10% FBS and antibiotics and 786-O cells were maintained in DMEM plus 10% FBS and antibiotics. All cells were grown at 37°C in a 100% humidified incubator with 5% CO2. Cells were grown to 80–90% confluency, harvested with trypsin, and resuspended to the cell density required for each assay.

In vitro angiogenesis (matrigel tube formation) assay

The matrigel tube formation assay was performed as previously described.22 Different treatments were added to the cells before plating onto the matrigel-coated plates. After 12–16 hr of incubation, tube formation was observed through an inverted photomicroscope (Nikon, Melville, NY). Microphotographs were taken and quantitative analysis was performed as previously described.22

Cell proliferation assay

Proliferation assays were performed as previously described.22 Briefly, 4 × 103 cells, suspended in endothelial basal medium supplemented with 1% FBS, were seeded into each well of a 96-well microtiter plate and treated with proliferation stimulants and active agents and incubated at 37°C for 72 hr. At the end of this period, WST-1 (10 μl) was added to each well, incubated at 37°C for 3 hr and absorbance at 450 nm was determined using a microplate reader (Bio-Rad, Hercules, CA). The experiments were performed in triplicate, and the figures presented represent the average of triplicate experiments.

Distribution of paclitaxel into cultured cells

The distribution of paclitaxel into cultured cells was determined as previously described23, 24 with modifications. Briefly, cells (1 × 105) were plated on each of 12-well plates and incubated overnight. The next day, cells were treated with 3H paclitaxel and incubated for 4 hr at 37°C. After this period, cells were washed on ice with PBS 3 times and then lysed with 1.5 M NaOH. The lysate was then added to 5 ml of scintillation fluid and radioactivity was measured for 5 min. Cells treated with cold (nonradiolabeled) paclitaxel, incubated at the same time and under the same conditions were used to determine the cell number for each cell type. Total cell count per well was determined. Total cell volume was determined by multiplying the number of cells by the volume of an average cell, as estimated by measuring the average diameter of 30 cells under a microscope with ×40 magnification. Single cell volume was calculated using the formula for the volume of a sphere. This information was used to calculate the radioactive equivalent concentration of paclitaxel within the cells. Experiments were performed in triplicate wells and repeated at least twice.

Determination of endothelial Cox-2

HUVE (2 × 106) resuspended in EBM-2 medium with 5% FBS and antibiotics and plated in 60 mm plates were treated with paclitaxel and incubated at 37°C for different time periods. At each time period, the cells were lysed with RIPA buffer (Boston Bioproducts, Ashland, MA) and protease inhibitor cocktail tablets (Roche Diagnostics, Indianapolis, IN). Protein concentration of the cell lysates was determined by the BCA™ protein assay (Pierce, Rockford, IL), following the manufacturer's recommendations. Cox-2 determination of the lysate (100 μL) was performed using a commercially available ELISA kit (Alpha Diagnostic International, San Antonio, TX), following the manufacturer's recommendations. Cox-2 concentration of the lysates was corrected for total protein concentration and presented as the average of triplicate measurements and standard deviations.

Patients and treatment plan

Protocols to perform pilot phase II studies of low dose, continuous i.v. infusion paclitaxel and oral celecoxib for patients with hormone refractory prostate cancer and melanoma were reviewed and approved by the local Institutional Review Board. Results of these studies will be formally reported elsewhere, when completed. Four patients (all the patients available at the time of this report) are presented here, and the data presented on the patients is focused on the translational studies. All patients signed a written informed consent document as a condition of participation in the study. Patients who failed standard therapy and had preserved organ function were eligible. The minimum time from any prior treatment and entry into these protocols was 1 month. The dose of paclitaxel was based on a phase I trial of low dose, continuous infusion paclitaxel in combination with radiation therapy.25 Paclitaxel was delivered through a central catheter as a continuous i.v. infusion at a rate of 10 mg/m2/day for 6 consecutive days without interruption, followed by 1 day without treatment. Administration of celecoxib 400 mg p.o. twice a day at 7:00–8:00 in the morning and evening was initiated after completing the first week of paclitaxel. Treatment was continued until disease progression or unacceptable toxicity occurred.

Blood collection

Blood samples for biological assays were drawn from a peripheral vein in the arm contralateral to the site of drug administration and collected directly into citrated tubes. Sample tubes were mixed by inversion and placed on wet ice until centrifuged (3,210g for 10 min, 4°C). The plasma was sterilized by filtration through a 0.2 μm membrane filter (Millipore Corporation, Bedford, MA) and then stored at −70°C until assayed. Lepirudin (Aventis Pharmaceuticals, Kansas City, MO) at a concentration of 10 U/ml was added to plasma before the cell based bioassays to prevent clot formation. Serum was obtained from blood collected in serum isolator tubes. After allowing the blood to stand at room temperature for 30 min to induce clot formation, the tubes were centrifuged at 3,210g for 10 min and the serum filter-sterilized prior to storage at −70°C.

Blood specimens (7 ml) for drug concentration determinations were drawn from a peripheral vein in the arm contralateral to the site of drug administration and collected directly in tubes containing freeze-dried sodium heparin anticoagulant. Samples were obtained before dosing, once a week during the first month of therapy, and every other week thereafter. Sample collection was restricted to days 4–6 of each weekly infusion of paclitaxel, at least 1 hr before or after replacing the dosing solution, to insure that steady-state conditions had been achieved. In practice, the samples were obtained 6 to 8 hr after the morning dose of celecoxib was taken. Sample tubes were mixed by inversion and placed on wet ice until centrifuged (1,000g, 10 min, 4°C) within 15 min. Plasma was transferred into a polypropylene cryovial and stored at −70°C until assayed. Frozen plasma samples were thawed at room temperature, mixed on a vortex stirrer and centrifuged for 5 min at 10,000g prior to drug concentration analyses.

Determination of serum vascular endothelial growth factor (VEGF), interleukin-8 (IL-8) and basic fibroblast growth factor (bFGF)

Serum concentrations of angiogenic growth factors VEGF, IL-8 and bFGF were measured from patient samples (in triplicate) at different time points during therapy. Measurements were performed with commercially available ELISA kits for VEGF 165 (product number DVE00), IL-8 (product number D8050) and bFGF (product number DFB50), obtained from R&D systems (Minneapolis, MN), following the manufacturer's specifications, using controls and standards included in the kit.

Determination of paclitaxel in plasma samples

Paclitaxel (purity ≥ 97%) purchased from the Sigma Chemical Co. (St. Louis, MO) was used to prepare calibration standards and quality control samples of the drug in human plasma. Plasma samples (500 μl) were prepared for analysis by extraction with 3 ml of tert-butyl methyl ether after adding 15 μl of a 300 ng/ml solution of cephalomannine (LKT Laboratories, St. Paul MN) in N,N-dimethylformamide as an internal standard. The organic solvent was removed, evaporated, and the extract reconstituted with 75 μl of N,N-dimethylformamide and 75 μl of 10 mM ammonium formate buffer, pH 3.75. The sample extract (100 μl) was loaded onto a 15 cm × 4.6 mm Luna 3 μm C18(2) HPLC column (Phenomenex, Torrance, CA), preceded by a 4 mm × 3 mm precolumn of the same stationary phase. The column was eluted at ambient temperature with acetonitrile-10 mM ammonium formate buffer, pH 3.75 (54:46, v/v) at a flow rate of 1.0 ml/min. Electrospray ionization mass spectrometric detection was performed using a Hewlett Packard (Palo Alto, CA) 1100 Series LC/MSD system. Operating parameters of the atmospheric pressure ionization-electrospray interface were the following: nebulizer pressure, 50 p.s.i.; drying-gas, N2; drying-gas flow, 10 liter/min; drying-gas temperature, 250°C and capillary potential, 5,000 V. The single-quadrupole mass spectrometer was operated in the selected-ion monitoring mode to measure positive ions corresponding to the protonated molecular ions of paclitaxel and cephalomannine at m/z 854.3 and 832.4, respectively, with a mass width of 0.6–0.7 μm amu; dwell time of 289 msec and a 75 V fragmentor potential. The concentration range of the calibration standards of paclitaxel in human plasma was 0.50 to 99.0 ng/ml. Accuracy of the assay for measuring 5 independently prepared sets of quality control samples with added paclitaxel concentrations of 2.23, 45.6 and 89.1 ng/ml ranged from 92.4 to 94.9% and the precision was 2.8 to 4.3%. Accuracy and precision for measuring paclitaxel at the 0.5 ng/ml lower limit of quantitation were 108.7% and 6.6%, respectively.

Determination of celecoxib in plasma samples

Celecoxib (purity, 99.28%) purchased from LKT Laboratories was used as the analytical reference sample for the assay. The internal standard, N,N-dimethylcelecoxib, was synthesized from the reaction of celecoxib with excess iodomethane in a 2-phase mixture composed of methylenechloride and a solution of tetrabutylammonium hydrogen sulfate in 0.1 M sodium hydroxide with vigorous shaking for 1 hr at room temperature. The desired product was isolated in excellent purity and yield (99.2%) upon evaporation of the organic phase and used without further purification.

Protein was removed from plasma samples (50 μl) by precipitation induced by adding 150 μl of a 200 ng/ml solution of the internal standard in acetonitrile and vigorously mixing. After centrifuging for 5 min at 10,000g, 10 μl of the clear supernatant was directly injected onto a 15 cm × 4.6 mm Luna 3 μm C18(2) HPLC column (Phenomenex, Torrance, CA), preceded by a 4 mm × 3 mm precolumn of the same stationary phase. An isocratic mobile phase composed of acetonitrile-1% acetic acid (80:20, v/v) delivered at 1.0 ml/min was used for the separation at ambient temperature. Elution of the drug and internal was detected by electrospray ionization mass spectrometry using the same instrument described in the above. Operating parameters of the interface were the following: nebulizer pressure, 30 p.s.i.; drying-gas, N2; drying-gas flow, 12 liter/min; drying-gas temperature, 350°C and capillary potential, 5,000 V. Protonated molecular ions of celecoxib and internal standard at m/z 382.1 and 410.1, respectively, were measured in the positive mode by selected-ion monitoring (mass width, 0.6–0.7 amu; dwell time, 289 msec; fragmentor potential, 125 V). The concentration range of the calibration standards of celecoxib in human plasma was 27 to 2,660 ng/ml. Accuracy of the assay for measuring 5 independently prepared sets of quality control samples with added celecoxib concentrations of 80, 789 and 2,394 ng/ml ranged from 98.4 to 103.2% and the precision was 1.63 to 3.16%. Accuracy and precision for measuring celecoxib at the 27 ng/ml lower limit of quantitation were 101.2% and 1.24%, respectively.

Quantitation

Chromatograms were integrated to provide peak areas using the data analysis functions of the Hewlett Packard ChemStation software. Study samples were independently assayed in duplicate, on separate days, together with a series of calibration standards and set of 3 quality control samples of the drug in human plasma. Standard curves were constructed by plotting the drug:internal standard chromatographic peak area ratio against the known concentration of drug in the calibration standards. Linear least squares regression was performed using a weighting factor of 1/y. Values of the slope and y-intercept of the best fit line were used to calculate the drug concentration in study samples. Specimens with concentrations exceeding the upper range of the standard curve were reassayed upon dilution with drug-free human plasma. The average of the 2 determinations of each study sample was calculated. Samples were reassayed in cases where the individual determinations differed from their average by more than 10%.

Data analysis

The steady-state concentration in plasma (Css) of paclitaxel was calculated as the geometric mean of all determinations made during the infusion. The mean concentration of celecoxib in plasma 6 to 8 hr after dosing at steady state [Css(6–8 hr)] for the twice daily regimen was similarly calculated.

Statistical analysis

The results of the in vitro assays are expressed are presented as means ± SD, unless otherwise specified. Means were compared by use of a Student's t test analysis. Differences were considered statistically significant at p<0.05. All statistical tests were 2-sided.

Results

Low concentrations of paclitaxel inhibit in vitro angiogenesis

Proliferation and tube formation assays on 3 different human EC types (HUVEC, HMVEC-L and HMVEC-D) were performed. A significant dose-dependent inhibition of bFGF- and VEGF-induced proliferation at 72 hr was observed. At a concentration of 5 nM of paclitaxel, VEGF-induced proliferation was inhibited by 54.5% and bFGF-induced proliferation was inhibited by 63%, in HUVEC (Fig. 1a), by 49.8% in HMVEC-L (Fig. 1b), and by 67.5% and in HMVEC-D, respectively (p < 0.01, 2-sided Student's t test, Fig. 1c). Figure 2 shows the effects of paclitaxel on tube formation on the 3 EC types. A similar dose response was observed and significant inhibition (p < 0.05) was seen at paclitaxel concentrations between 5 and 10 nM in all 3 EC types (Fig. 2a, and representative pictures in Figure 2b).

Figure 1.

Effects of paclitaxel on EC proliferation. ECs were treated with increasing concentrations of paclitaxel and proliferation was measured at 72 hr as explained in materials and methods. A dose dependent inhibition of proliferation in HUVEC (a), HMVEC-L (b) and HMVEC-D (c) was observed. Potent inhibition of proliferation was observed at paclitaxel concentrations of 5 nM in all 3 endothelial cell types. Results are the average of triplicate experiments, expressed as percentage of control, ± SD. p ≤ 0.01 (control vs. 5 nM paclitaxel for all EC types).

Figure 2.

Effects of paclitaxel on EC tube formation. (a) Dose response curve of paclitaxel's effects on tube formation on HUVEC, HMVEC-L and HMVEC-D; p < 0.05 (5 and 10 nM paclitaxel, compared to untreated control, for all 3 EC types). (b) Representative pictures (×20) of the inhibitory effects of paclitaxel at 5 nM in macro and microvascular endothelial cells are presented. Left column: HUVEC; middle column: HMVEC-L; right column: HMVEC-D. Upper row: untreated endothelial cells. Lower row: endothelial cells treated with 5 nM of paclitaxel. Bars: 500 μm. Results are the average of triplicate experiments, expressed as percentage of control, ± SD.

In order to rule out the possibility that the endothelial inhibitory effects of paclitaxel were attributable to the vehicle component Cremophor EL, experiments using equivalent concentrations of Cremophor EL as contained at the concentrations of paclitaxel used in the above experiments were performed. No inhibitory effects on endothelial proliferation or tube formation (data not shown) were observed. Experiments using commercially available semisynthetic paclitaxel (Calbiochem, San Diego, CA) diluted in DMSO/PBS showed potent antiendothelial effects as paclitaxel from the clinical dosage form (data not shown).

Human endothelial cells accumulate paclitaxel more than nonendothelial cells

Given the increased sensitivity of endothelial cells to low nanomolar concentrations of paclitaxel, it was hypothesized that endothelial cells may have an increased ability to accumulate paclitaxel over nonendothelial cells. Intracellular radioactivity was measured 4 hr after treating HUVEC, HMVEC-D, and HMVEC-L with 5 nM of [3H]-paclitaxel, a concentration that exhibits significant endothelial cell inhibition, and compared to the human nonendothelial cells IMR-90 (primary fibroblasts), PC-3 (prostate cancer) and 786-0 (renal cell cancer). The effects of low nanomolar concentrations of paclitaxel (up to 10 nM) on proliferation of the above nonendothelial cell types were modest (Fig. 3).

Figure 3.

Effects of low concentrations of paclitaxel on proliferation of nonendothelial cells. Primary human fibroblasts (IMR-90), human prostate cancer (PC-3) and human renal cell carcinoma (786-0) cell lines were treated with paclitaxel (0.1–10 nM) and proliferation was measured at 72 hr as in Material and Methods. A modest inhibition in proliferation was achieved in IMR-90 (white bars), PC-3 (black bars) and 786-0 (gray bars) cells at concentrations up to 10 nM. Values are the average of triplicate experiments, expressed as percentage of control, ± SD. p: 0.19 for IMR-90; 0.03 for PC3, 0.2 for 786-0 (10 nM paclitaxel compared to control).

Intracellular levels of radioactivity in the 3 endothelial cell types were higher than in non-ECs, resulting in higher apparent paclitaxel concentrations in ECs (Table I). These findings suggested that the accumulation of paclitaxel in ECs was more than 5 times greater than in non-ECs.

Table I. Apparent Paclitaxel Concentrations in Endothelial and Nonendothelial Cells1
Cell typePaclitaxel radioactive equivalent concentrations (NM) +/− SEAccumulation factor
  • 1

    Concentration of [3H] paclitaxel added to the cells was 5 nM.

HUVEC678 +/− 8.3136
HMVEC-D537 +/− 5.7107
HMVEC-L942 +/− 109.1188
IMR-9087 +/− 7.317
PC-3140 +/− 6.928
786-0177 +/− 49.335

Low-dose paclitaxel induces Cox-2 expression in endothelial cells

It has been shown that paclitaxel increases the expression of cyclooxygenase-2 (Cox-2) in mononuclear cells, tumor cells, and in mammary epithelial cells.21, 26, 27 In these studies, the concentrations of paclitaxel required to induce Cox-2 expression were in the micromolar range. Since endothelial cells accumulate paclitaxel to a greater degree than nonendothelial cells, the effects of nanomolar concentrations of paclitaxel on Cox-2 expression in endothelial cells were investigated. HUVE cells were treated with paclitaxel (Sigma-Aldrich Research, St. Louis, MO) for 8 and 24 hr and total cell lysates were obtained for determination of Cox-2 by ELISA. As shown in Figure 4, increased levels of Cox-2 protein were observed in HUVEC after 24 hr treatment with 1, 10 and 50 nM of paclitaxel, compared to baseline. Increased levels of endothelial COX-2 were induced at 8 hr by 50 nM of paclitaxel.

Figure 4.

Induction of Cox-2 protein expression by low doses of paclitaxel. HUVE cells, untreated and treated with paclitaxel at 1, 10 and 50 nM in endothelial cell medium and 5% FBS were incubated at 37°C for 8 and 24 hr. At the end of these time periods, cells were lysed and ELISA assay for Cox-2 was performed. At 24 hr of treatment, an increase of Cox-2 protein levels was observed at all paclitaxel concentrations. At 8 hr, increased levels of COX-2 were observed at concentrations of 50 nM. COX-2 concentrations are presented as ng/mg of total protein.

In vitro antiangiogenic effects of paclitaxel are enhanced by Cox-2 inhibition

Based on the above findings, it was hypothesized that the antiangiogenic effects of paclitaxel could be further enhanced by simultaneous treatment with Cox-2 inhibitors. HUVE cells were treated with the Cox-2 inhibitor, NS-398 and paclitaxel, both alone and in combination, and endothelial cell proliferation was assessed (Fig. 5). At a concentration of 1 μM, NS-398 inhibited HUVEC proliferation by 13.8%, and at 10 μM the degree of inhibition increased to 40%. Paclitaxel at a concentration of 1 nM inhibited HUVEC proliferation by 28.3%. When added simultaneously, a more pronounced inhibitory effect on HUVEC proliferation induced by bFGF (71.7% inhibition) was seen at paclitaxel concentrations of 1 nM and NS-398 concentrations of 1 μM (Fig. 5).

Figure 5.

Effects of combined treatment with low doses of paclitaxel and Cox-2 inhibitors. HUVE cells were treated with low concentrations of paclitaxel and the Cox-2 inhibitor NS-398, both alone and in combination, and bFGF-induced proliferation was measured at 72 hr. NS 398 potently inhibited HUVEC proliferation at 10 μM, but not at 1 μM. Paclitaxel had a mild inhibitory effect at 1 nM and very potent effects at 5 nM. The inhibitory effect was potently enhanced when both agents were simultaneously added (1 nM of paclitaxel and 1 μM of NS 398). Unstim and Stimul refer to untreated HUVEC controls, unstimulated and stimulated. TX: paclitaxel; NS: NS-398; T1, N1, T1N9, T1N10: combined treatment with paclitaxel (mamomolar) and NS-398 (micromolar). Values are the average of triplicate experiments, expressed as percentage of control (untreated endothelial cells) ± SD. p: 0.01 (combination of paclitaxel 1 nM and NS398 1 μM vs. paclitaxel at 1 nM alone).

Systemic administration of low dose paclitaxel and celecoxib induce antiangiogenic activity in the plasma of cancer patients

In order to extend the laboratory findings into a potentially feasible therapeutic strategy for patients with cancer, we started pilot clinical trials of low dose, continuous infusion paclitaxel and celecoxib in patients with melanoma and prostate cancer. The trials are currently underway, and the results will be formally reported elsewhere, when the accrual goals are reached. Here, we present data on 4 cancer patients (all patients accrued by the time of the article submission), focusing on pharmacokinetic and pharmacodynamic studies, as well as early evidence of potential benefit. Four cancer patients (patient 1: metastatic melanoma to lungs and mediastinum; patient 2: ocular melanoma with diffuse liver metastases; patient 3: hormone refractory prostate cancer and patient 4: hormone refractory prostate cancer) were treated with continuous infusion paclitaxel and celecoxib in a pilot clinical trial, as described in Material and Methods. All patients had documented disease progression and signed informed consent. Patients were initially treated with paclitaxel (week 1), and celecoxib was started on week 2. Blood was obtained at baseline and at different time points during treatment, for pharmacokinetic analysis, angiogenesis bioassays and cytokine determinations.

As shown in Figures 6 and 7, a significant antiangiogenic effect (as assessed by the tube formation and proliferation assays) was detected in the plasma of patients after starting therapy (baseline vs. 4 weeks p: 0.0003 for tube formation; p: 0.004 for proliferation). Interestingly, the antiangiogenic effects of the patients' plasma were more pronounced when patients were treated with the combination therapy (paclitaxel and celecoxib, weeks 4 and later), compared to paclitaxel alone (week 1 of therapy), similar to our in vitro findings. Steady state plasma levels of paclitaxel were around 10 nM, and levels of celecoxib were between 1 and 5 μM (Table II).

Figure 6.

Effects of patients' plasma before and during treatment on endothelial tube formation. Patients were treated as in materials and methods; plasma was obtained at different time points during treatment and used for tube formation assays. Cell cultures contained 10% (v/v) of patient plasma in endothelial cell medium supplemented with 1% FBS. (a) Compared to baseline, plasma obtained during treatment potently inhibited tube formation (p:0.0003, baseline -0 weeks- vs. week 4; p: 0.0002, baseline vs. 12 weeks). The inhibitory effects were more pronounced when patients were treated with the combination (paclitaxel and celecoxib, weeks 4, 8 and 12) compared to paclitaxel alone (week 1). closed circle, Patient 1; open square, Patient 2; *, Patient 3; open triangle, Patient 4. Results are presented as percentage of baseline and each data point represents the average of experiments performed in triplicate. (be) Representative microphotographs (×40 magnification) of the anti-tube formation effects of the plasma of patient 3 as a result of treatment. (b) Baseline plasma (before treatment) stimulated endothelial tube formation. (c) Tube formation was decreased by plasma taken during week 1 of treatment (paclitaxel alone). (de) The plasma inhibitory effects were more marked during combined therapy with paclitaxel and celecoxib, compared to paclitaxel alone.

Figure 7.

Enhanced antiproliferative effects of the plasma of patients during treatment with paclitaxel and celecoxib. Using patients' plasma at baseline and at different time point during treatment, 72 hr proliferation assays were performed on HUVEC. Cell cultures contained 10% (v/v) of patient plasma in endothelial cell medium supplemented with 1% FBS. Inhibitory effects were observed on endothelial proliferation p: 0.004, baseline vs. week 4; P: 0.001, baseline vs. week 8), and were more marked when patients were receiving the combination (paclitaxel and celecoxib, weeks 4 and later), as compared to paclitaxel alone (week 1). closed circle, Patient 1; open square, Patient 2; *, Patient 3; open triangle, Patient 4. Results are presented as percentage of baseline and each data point represents the average of experiments performed in triplicate.

Table II. Steady-State Concentrations of Paclitaxel and Celecoxib in Patients Treated With a Protracted Continuous I.V Infusion of 10 MG/M2/Day Paclitaxel And 400 MG Celecoxib P.O. Every 12 hr
Patient numberPaclitaxelCelecoxib Cmath image (μM)
CSS (nM)CL (liters/h/m2)
112.1 ± 3.541.22.9 ± 1.9
28.3 ± 0.958.81.3 ± 0.9
315.2 ± 3.532.12.3 ± 1.2
414.9 ± 0.132.84.9 ± 0.7

Of the 4 patients treated, there was evidence suggestive of clinical benefit in 3 of them. Patient 2, with rapidly progressive ocular melanoma metastatic to the liver, refractory to biologic response modifiers and chemotherapy, had stable disease and improvement in his performance status for approximately 15 weeks, before disease progression. Patient 3, with hormone refractory prostate cancer, urinary obstruction requiring nephrostomy tubes and PSA doubling, had symptomatic improvement in the obstruction and more than 50% reduction in serum PSA during treatment (from 31.8 ng/ml at baseline to 26.7 ng/ml by week 4 and 14 ng/ml by week 12). Representative pictures of the plasma inhibitory effects on tube formation are presented in Figure 6be. The patient voluntarily withdrew from the study after 12 weeks of therapy because of catheter malfunction, requiring replacement. Patient 4, with hormone refractory prostate cancer and rapid doubling of his serum PSA over the last month prior to therapy (99.7 ng/ml at week (−) 4 to 187 at week 0), had a decrease in the rate of PSA progression (202 ng/ml at week 4 and 216 ng/ml at week 6).

Serum levels of VEGF and IL-8 in the 4 patients were highly variable and not correlated with the angiogenic activity in the plasma, plasma levels of paclitaxel and celecoxib, or clinical activity (data not shown). Serum levels of bFGF were below the detection limits of the kit.

Discussion

Our study shows several novel findings: first, human ECs can accumulate paclitaxel more than 5 times greater than normal human fibroblasts and several human cancer cell lines. This may offer an explanation to preclinical reports describing ECs' exquisite in vitro sensitivity to paclitaxel, the in vivo antiangiogenic effects of the drug, and clinical reports of the relative efficacy of low weekly doses of paclitaxel in patients refractory to standard (every 21 days) doses of this agent.28, 29, 30, 31 Whether the ability of ECs to accumulate paclitaxel is due to increased transport of drug into the cells or to decreased drug efflux is not known.

Our findings that low nanomolar concentrations of paclitaxel inhibit angiogenesis in vitro confirm previous reports.3, 15, 16 These findings were further extended to human microvascular endothelial cells from lung (HMVEC-L) and dermis (HMVEC-D) and demonstrate similar sensitivity to paclitaxel as HUVEC, both in tube formation and in proliferation. Moreover, HMVEC-L and HMVEC-D also accumulated paclitaxel to a similar extent as HUVEC. Since lungs and skin are common sites of metastases from human malignancies, the demonstration that ECs derived from lung and dermal microvasculature are equally sensitive to low concentrations of the drug may have important biological implications.

Even though the ability of ECs to accumulate paclitaxel may be important to the antiangiogenic effects exhibited by the drug, additional pharmacological factors that are unrelated to this finding may be involved. Paclitaxel at low nanomolar concentrations can also affect tumor angiogenesis indirectly either by reducing the tumor secretion of angiogenic growth factors, such as VEGF (10) or by decreasing its secretion by endothelial cells, which may act in an autocrine fashion (Merchán and Sukhatme, unpublished findings).5 Other possible mechanisms that may explain the antiangiogenic effects of this combination include a decrease in circulating endothelial precursor number and viability, as previously demonstrated with metronomic cyclophosphamide.32 Other tumor cells are known to significantly accumulate paclitaxel as well.23, 24, 33, 34 Our results can not be directly compared to those previously reported because of the different cell types and different methods used.

Second, this report shows that addition of Cox-2 inhibitors to low nanomolar concentrations of paclitaxel enhances its antiangiogenic effects. Increased Cox-2 expression was demonstrated by ELISA in HUVE cells treated with nanomolar concentrations of paclitaxel at 24 hr (Fig. 4). Higher (micromolar) concentrations of paclitaxel also induce Cox-2 expression in epithelial and cancer cells.21, 27 The mechanisms by which taxanes enhance COX-2 expression have been described and may involve both stimulation of transcription and RNA stability.35 This may be one mechanism by which standard doses of paclitaxel that achieve peak plasma concentrations in the micromolar range may be associated with inflammatory reactions such as arthralgias, when given to patients with cancer.21, 27, 36 Induction of Cox-2 in tumor cells and in endothelial cells by paclitaxel may be a “protective” mechanism counterproductive to the overall antiangiogenic and antitumor effects of paclitaxel. This study suggests that even when paclitaxel is given according to a protracted low dose schedule (metronomic dosing), it still may increase Cox-2 production by the endothelium. The finding that a Cox-2 inhibitor (i.e., NS-398) potentiated the effects of paclitaxel on HUVEC proliferation provides a rationale for evaluating the use of paclitaxel at low doses concomitantly with a Cox-2 inhibitor to enhance its antiangiogenic and possibly its antitumor effects.

Finally, this report demonstrates for the first time that treatment with paclitaxel as a protracted low dose continuous i.v. infusion together with twice daily oral administration of the Cox-2 inhibitor celecoxib is feasible and may be associated with biological and clinical activity. This therapy was associated with a clinical response in 1 patient, and disease stabilization in 2 others, with no significant adverse events, such as myelosuppression, neurotoxicity or alopecia. The steady-state plasma concentrations of paclitaxel and celecoxib achieved during treatment were close to the targeted levels in the low nanomolar range for paclitaxel and between 1 and 5 micromolar for celecoxib.

Our in vitro findings that the combination of paclitaxel and Cox-2 inhibition exerted an enhanced antiangiogenic effect in HUVEC compared to each agent alone were confirmed in the in vivo setting, when 4 patients with cancer were treated with paclitaxel and celecoxib. It was observed that the antiangiogenic effects of the patients' plasma (both in the tube formation and proliferation assays) were enhanced when the patients were exposed to the combination of paclitaxel and celecoxib (week 4 and later), compared to exposure to paclitaxel alone (week 1, Fig. 6a,b, 7). Moreover, the inhibition of endothelial functions by plasma specimens acquired from patients during treatment was gradual, tending to increase over time (Figs. 6, 7). This would suggest that the observed effects on in vitro angiogenesis were not entirely due to the presence of paclitaxel and/or celecoxib in the plasma but possibly due to changes in systemically circulating pro- and anti-angiogenic factors as a result of the treatment, the net effect of which was detected by our bioassays. The endothelial inhibitory effects seen in the patients' plasma (Figs. 6, 7) were less marked than the inhibitory effects observed with pure paclitaxel in the absence of plasma (Figs. 1, 2, 5). This could be explained in part by the fact that paclitaxel in plasma highly protein bound,37 which will reduce the effective concentration of the drug in the assay system. In addition, plasma is rich in proangiogenic factors that may decrease the inhibitory effects seen when paclitaxel is used alone in the absence of human plasma.

The measurement of angiogenic activity in serum or plasma specimens has been previously described by us and others as a potentially informative biomarker for monitoring the biologic effects of antiangiogenic therapies and the detection of angiogenesis dependent diseases.22, 38, 39 Here, it is shown that plasma angiogenic activity can be followed over time in patients treated with a metronomic regimen of paclitaxel and standard regimen celecoxib. A distinct association between the angiogenic activity in plasma of treated patients and clinical benefit was also suggested (Fig. 6). Patient 2 had stable disease for 16 weeks, and his plasma showed inhibitory effects during the first 12 weeks during therapy. Patient 3 demonstrated a clear correlation of progressive antiangiogenic effects of the plasma during treatment (Figs. 6, 7), PSA response and symptomatic improvement. Plasma obtained from patient 4, who had slowing of serum PSA doubling while on therapy, showed inhibitory effects on endothelial tube formation. These results provide preliminary evidence to suggest that plasma angiogenesis bioassays may be useful tools for monitoring the biologic effects of antiangiogenic therapies, especially when combination therapies that affect multiple targets are used and where measurement of those targets is technically difficult or not practical (i.e., multiple biopsies, etc). However, studies with larger number of patients in prospective clinical trials will be necessary in order to validate these assays as useful monitoring tools.

A clear association between the patients' plasma angiogenic activities during treatment and levels of serum angiogenic cytokines was not found. Serum levels of VEGF and IL-8 were variable, even when the disease was stable and the plasma showed antiangiogenic activity. These findings are in concordance with previous clinical trials of antiangiogenic agents, such as thalidomide, endostatin and metronomic chemotherapy, where serum angiogenic cytokines have not proven to be a reliable tool to monitor the biologic effects of antiangiogenic therapies over time.40, 41, 42, 43

An important unanswered question that derives from this work is whether or not the observed effects of low dose paclitaxel and oral celecoxib are solely due to antiangiogenesis or also due to tumor cytotoxicity from prolonged exposure of the tumor to low concentrations of paclitaxel and celecoxib. This possibility cannot be ruled out, but at the doses and plasma concentrations achieved, tumor cytotoxicity may not play the most important role. Cytotoxic concentrations of paclitaxel and, in general, of chemotherapeutic agents are associated with significant myelotoxicity and other side effects such as hair loss, neuropathy, etc. In the patients treated with this regimen, we did not see any such side effects. In vivo studies of drug penetration into tumors (and into tumor endothelium) after prolonged exposure to low doses of paclitaxel are needed in order to answer this important question. Additionally, since many human cancers overexpress the multi-drug resistance (MDR) genes as a mechanism of drug resistance, low doses of paclitaxel may not have any significant effect on such tumor cells but may have an important inhibitory effect on MDR negative endothelial cells, therefore overcoming resistance.

In summary, the antiangiogenic effects of low doses of paclitaxel may be in part due to the enhanced ability of human endothelial cells to accumulate the drug. “Antiangiogenic” concentrations of paclitaxel induce endothelial Cox-2 production, and Cox-2 inhibition potently enhances paclitaxel's inhibitory effects on endothelial cells. Combined treatment with paclitaxel delivered in a metronomic manner in combination with the Cox-2 inhibitor celecoxib was well tolerated in 4 patients with advanced cancer, induced a change in the angiogenic activity of the patients' plasma, and appeared to be associated with clinical benefit. Further studies to characterize the mechanisms of these effects and clinical trials formally testing this combination are currently underway.

Acknowledgements

We thank J. Chung for technical assistance. We also thank members of the Sukhatme laboratory for helpful discussion. This work was supported in part by seed funds from Beth Israel Deaconess Medical Center and the family of Victor J. Aresty, the Trust Family Foundation (to V.P.S.), the 2001 American Society of Clinical Oncology Young Investigator Award, the Clinical Investigator Training Program Fellowship (to J.R.M.), the Hershey Family Foundation for Prostate Cancer (to G.J.B.), and by the National Institutes of Health/ National Cancer Institute P30-CA0516 grant (to J.G.S.).

Ancillary