The influence of 5-fluorouracil and methotrexate on vascular endothelium. An experimental study using endothelial cells in the culture

Annals of Oncology 7: 731-737, 1996. O 1996 Kluwer Academic Publishers. Printed in the Netherlands.

Original article The influence of 5-fluorouracil and methotrexate on vascular endothelium. An experimental study using endothelial cells in the culture M. Cwikiel,1 J. Eskilsson,2 M. Albertsson1 & L. Stavenow3 Departments of' Oncology and 2 Cardiology, University Hospital, Lund; ^Department of Internal Medicine, Section for Vascular and Renal Diseases, University Hospital, Malmo, Sweden

Summary

Key words: cardiotoxicity, cell culture model, endothelium, 5-fluorouracil, prostacyclin

study in rabbits with electron microscopic evaluation of arterial endothelium after treatment with 5-FU [8]. The 5-Fluorouracil (5-FU), an anti-metabolite of the py- study showed very serious damage to endothelial cells rimidine analogue type, is one of the most commonly (EC), occasionally followed by platelet accumulation used chemotherapeutic drugs in the treatment of and fibrin formation, suggesting that thrombogenicity human malignancies. The well known adverse effects of could be attributed to 5-FU's direct toxicity. However, 5-FU are myelosupression, mucositis of the gastro- despite the severe damage, thrombus formations were intestinal tract and thrombophlebitis of peripheral relatively poorly represented in our material, which veins. Cardiotoxicity, its relatively unknown side effect, could be interpreted to show that endothelium-derived, is often a limiting factor of 5-FU's usefulness in clinical thrombus-preventing substances might be involved in oncological practice. the process. Methotrexate (MTX), a folic acid analogue antiThe endothelium plays an important role in preventmetabolite, is an important chemotherapeutic drug ing thrombus formation and participates in vasoreguwith an established role in cancer treatment. It can also latory mechanisms [9-11]. The results from previous be used in some non-malignant conditions such as studies on cell cultures, where toxic effects on various psoriasis or rheumatoid arthritis. MTX lacks cardio- endothelial cell functions such as proliferation and toxicity in its side-effect panorama, a fact which makes prostacyclin release were seen [12], encouraged us also the drug interesting for our study. to use this model. The pathophysiology of 5-FU induced cardiotoxicThe aim of the study was to investigate, by using a ity is still unknown [1, 2]. Several hypotheses have been cell-culture model, the effect of 5-FU on DNA syntheput forword such as coronary artery spasm [3-5], auto- sis, cell death, and release of prostacyclin by vascular immune-mediated injury of the myocardium [6] and EC. For comparison, we carried out the same tests on thrombogenic effects [7]. Another possible mechanism MTX, an antimetabolite without known cardiotoxic is a direct toxic effect on the coronary endothelium, a properties. hypothesis which has been tested in an experimental Introduction

Downloaded from http://annonc.oxfordjournals.org/ by guest on June 6, 2016

Background: Cardiotoxicity still remains an unexplained toxic manifestation of 5-fluorouracil (5-FU). Clinical and experimental data suggest that endothelium of coronary arteries could be involved in the pathophysiologjcal mechanisms of the syndrome. In order to further explain 5-FU induced cardiotoxicity, we investigated the influence of this drug on endothelial cells (EC) in a cell culture model. Materials and methods: The influence of 5-FU on EC, with respect to DNA synthesis, cell death and release of prostacyclin by endothelial cells (EC) was studied. For comparison, we tested methotrexate (MTX), an antimetabolite without cardiotoxic properties, in the same way. Human endothelial cell lines (HEC) and bovine endothelial cells

(BEC) were incubated with increasing concentrations of 5-FU and MTX for 48 hours. (3H)thymidine incorporation, total cellular protein, loss of (3H)thymidine from prelabelled cells and 6-keto-prostaglandin Fl were measured. Results: DNA synthesis decreased significantly in both HEC and BEC, and the release of prostacyclin by BEC increased significantly when incubated with 5-FU. This effect was not seen with MTX. Conclusion: The results indicate specific susceptibility of benign EC to 5-FU. Such susceptibility was confirmed by the release of prostacyclin by the BEC, indicating leakage secondary to EC injury.

732

Materials and methods

Cell death

Endothelial cell lines were incubated with increasing concentrations of 5-FU and MTX for 48 hours. (3H)thymidine incorporation, total cellular protein, loss of (3H)thymidine from prelabelled cells and 6-keto-prostaglandin Fl were measured.

The loss of (3H)thymidine from prelabelled cell layers, a slight modification (23] of the method by Eckel & Fujimoto [24] was used. Cell count Cells were counted in a Burker hemocytometer chamber.

Chemicals 5-FU (Fluracedyl 50 mg/ml, Nycomed, Norway) and MTX (Methotrexate 2.5 mg/ml, Lederle, USA) were used. The concentrations of the drugs used were based on results from other in vitro studies on vascular EC [13, 14], and tested for dose dependency on human EC. Suitable concentrations were found to be: 5-FU, 5 \i/mi and 10 ^g/ml and MTX, 20 ng/ml and 40 ng/ml. These concentrations correspond to the high doses of 5-FU and conventional doses of MTX, normally used in clinical situations in oncological practice (15-18).

Culture of EC

Experimental procedures 2

For the experiments cells were trypsinized and seeded in 5.6 cm multiple wells (NUNC) at 0.6-0.8 x lOVcm2. Cells were allowed to establish growth for 24 hours and by this time they were approximately 50% confluent. The medium was then changed to fresh DMEM/Ham's F12 with 10% FBS (BEC) or MEM with 10% FBS but without HAT (HEC) and incubation with the drugs was started. Assay of (3H)thymidine incorporation During the last 4 h of the incubation time with the drugs, 0.033 \iG/ ml medium of methyl-(3H)thymidine (Amersham) was added to cultures. (3H)-activity in trichloroacetic acid resistant material was determined as described by [19] with a slight modification [20]. Assay of total cellular protein Protein was measured using Folin phenol reagent after alkaline copper treatment (Lowry method) [21]. Assay of prostacyclin A RIA method described in [22] was used to measure 6-ketoprostaglandin F l a (6-keto PGFla), a stable conversion product of prostaglandin I2 in the culture medium.

The Wilcoxon Matched-Pairs Signed Ranks Test was used.

Results Human endothelial cells It is difficult to avoid great variation in the absolute (3H)thymidine counts. Such variations may be explained by small differences in cell density or cell handling (subculturing, seeding) which, together with donor differences, may affect the metabolic activity and general condition of the cells. Therefore, in order to make comparisons between experiments possible, (3H)thymidine counts as well as total cellular protein were presented as a percentage of control values. In the experiments evaluating the effect of 5-FU on HEC, the absolute (3H)thymidine count in controls varied between 376 and 14898 d.p.m. per well, mean 6035 (SD 4147), reflecting basal cell proliferation. The absolute total cellular protein values in controls varied between 0.112 and 0.178 mg/well, mean 0.160 mg/well (SD 0.020). Effect of 5-FU on (3H)thymidine incorporation and total cellular protein of HEC in relation to time At 15 min incubation no effect of 5-FU on (3H)thymidine incorporation or total cellular protein was seen in the HEC. At 24 hours DNA synthesis in the HEC had decreased with increasing concentrations of 5-FU while it was not influenced by increasing concentrations of MTX. As shown in Figure 1, DNA synthesis in HEC at 48 hours incubation decreased with increasing concentrations of 5-FU. (3H)thymidine incorporation in EC incubated with 5-FU at 5 u.g/m] had decreased by 46% (P- 0.0047) and with 5-FU at 10 ng/ml by 60% (P0.0022). Compared to controls, incubation with 5-FU for 48 hours caused decreasing total cellular protein levels in the HEC, with increasing concentrations of 5-FU (Figure 2). Incubation with 5-FU at 5 u,g/ml caused an 11% decrease and with 5-FU at 10 jig/ml a 14%.

Downloaded from http://annonc.oxfordjournals.org/ by guest on June 6, 2016

Both a human endothelial cell line (HEC) and bovine endothelial cells (BEC) were used. Bovine EC were scraped from the aorta with a surgical blade into serum free medium and grown in Dulbecco's modified Eagle's medium (DMEM)/Ham's F12 in the ratio 1:5 containing 10% fetal bovine serum and gentamicine (0.1 mg/ml). The cells were seeded in gelatine-coated 25 cm2 plastic flasks (NUNC) and incubated at 37 *C in a humidified 5% CO2 atmosphere. Subcultures were obtained by the use of 0.05% trypsin-0.02% EDTA (Gibco). Confluent cells from passage two or three were used for the experiments. EC were identified by the unique square morphology and only EC with this typical appearance were used. The human endothelial cell line EAJiy926 [25] was cultured in Minimal Essential Medium (MEM) (Gibco) containing 10% FBS + HAT (1 ml/100 ml) (Hypoxanthine 5 mmol, Aminopterin 20 mmol, Thimidine 0.8 mmol) (Gibco). Subcultures were obtained as for BEC.

Statistical comparisons

733

Effect of 5-FU and MTX on the release ofprostacyclin by HEC

Per cent

uu

*•

-

p - 0.0022

80 -

Prostacyclin was not detected by our assay in the medium in which HEC had been grown. Bovine endothelial cells

p = 0.0047

60 -

Data are presented as a percentage (%) of control values in order to make the comparison between experiments possible. In the experiments with 5-FU, the absolute (3H)thymidine counts in controls varied between 519 and 5103 d.p.m. per well, mean 1947 (SD 1323) and the absolute total cellular protein values of controls varied between 0.194 and 0.315 mg/well, mean 0.258 mg/well (SD 0.040).

40 -

20 -

n

Effect of 5-FU on (3 Hjthymidine incorporation and total cellular protein in BEC incubated for 48 hours Figure 1. Effect of 5-FU on (3H)thymidine incorporation of culture Control

5 pg/m 5-FU

10

5-FU

Per cent •

00 -

p = 0.058

p = 0.068

80 -

60 -

DNA synthesis in the BEC had decreased significantly after 48 hours incubation with increasing concentrations of 5-FU, as shown in Figure 3. Incubation with 5-FU at 5 ng/ml inhibited (3H)thymidine incorporation in BEC by 95% (P •= 0.005) and with 5-FU at 10 Hg/mlby97%(P-0.005). In Figure 4, the effect of increasing concentrations of 5-FU on total cellular protein in cultured BEC, incubated for 48 hours is shown. The total cellular protein of the BEC decreased by 37% (P = 0.018) after 48 hours incubation with 5-FU at 5 fig/ml, and by 38% (P= 0.018) after 48 hours incubation with 5-FU at 10 Hg/ml.

\ Per cent

40 -

100 -

-> p = 0 005

20 80 n Control

5pq/ml5-FU

10|jq/ml5-FU

Figure 2. Effect of 5-FU on total cellular protein of HEC incubated for 48 hours. Confluent HEC were exposed to 5-FU at 5 or 10 ug/ ml. After 48 hours the total cellular protein was measured. The data represent percentage of control values and are derived from nine experiments with three wells in each experiment.

60 -

40 -

20 p = 0.005

Effect of MTX on (3 Hjthymidine incorporation in cultured HEC incubated for 48 hours At 48 hours the DNA synthesis in HEC was unchanged after being exposed to increasing concentrations of MTX.

Control

5 pg/ml 5 FU 10 ug/ml 5FU

Figure 3. Effect of 5-FU on (3H)thymidine incorporation in BEC incubated for 48 hours. BEC were subjected to 5-FU at 5 or 10 u.g/ ml and incubated for 48 hours. Each point is derived from ten separate experiments, and data in each experiment from three wells.

Downloaded from http://annonc.oxfordjournals.org/ by guest on June 6, 2016

HEC incubated for 48 hours. HEC were exposed to 5-FU at 5 u.g/ ml and 10 u.g/ml for 48 hours. Controls received no drug. The data are presented as percentages control values. Each point is derived from twelve separate experiments, and data in each experiment from three wells.

734 and 62.0 pg/ml, mean 22.3 pg/ml (SD 23.7) and the absolute 6 keto PGFla/total protein ratio varied between 24 and 492 ng/mg total protein, mean 221 ng/

Per cent

uu n- nf

80 -

Per cent

p = 0.018

p = 0.028

60 -

200 -

40 -

160 -

20 -

120 -

n

-

.525

p = 0.027

80 Control

5 pg/ml 5-FU

10 pg/ml

Control

5 uQ/ml 5-FU

10Mg/ml5-FU

nl5-FU

Figure 5A. Effect of 5-FU on the release of prostacyclin by BEC,

Effect of MTX on (3H)thymidine incorporation and total expressed as the absolute 6 keto PGFla/100 ul medium. BEC were incubated with 5-FU at 5 and 10 ng/ml for 48 hours. Each point is cellular protein in BEC incubated for 48 hours derived from seven separate experiments, and data in each experiment from three wells.

At 48 hours DNA synthesis in the BEC had increased with increasing concentrations of MTX, whi lular protein levels remained relatively unchi 320 -i

Ejfect of 5-FU on the release ofprostacyclin I The absolute 6 keto PGFla values of con 280 between 3.68 pg/ml and 56.5 pg/ml, mean (SD 21.4). The absolute 6 keto PGFla/U 240 ratio varied between 28 and 450 ng/mg to mean 196 ng/mg, (SD 191). Data are presei centage of control values. 200 As shown in Figures 5A and 5B, the prostacyclin increased both absolutely and total cellular protein during incubation with 160 ure 5A shows that 5-FU at a concentration increased the absolute 6 keto PGFla levi (P- 0.027). A 5-FU concentration of 1C 120 duced an increase of 98% (P - 0.028). Fron it may be seen that the release of prostacyc exposed to 5-FU at 5 ug/ml, expressed as a total cellular protein increased three-fold a controls ( P - 0.027). The release of pros BEC exposed to 5-FU at 10 ng/ml was twice that in controls. Effect of MTX on the release ofprostacyclin I

-

-

-

-

p = 0 027

-

Cc"'r'.'

In the experiments evaluating the effect of MTX on the release of prostacyclin by BEC, the absolute 6 keto PGFla levels of controls varied between 3.33 pg/ml

5 pq'ml 5-FU

Ljq.'m! 5-FU

I 5-FU

Figure SB. Effect of 5-FU on the release of prostacyclin by BEC, expressed as a fraction of total cellular protein. BEC were incubated with 5-FU at 5 and 10 (ig/ml for 48 hours. The experiments were performed six times and in three wells in each experiment

Downloaded from http://annonc.oxfordjournals.org/ by guest on June 6, 2016

Figure 4. Effect of increasing concentrations of 5-F1 40 lular protein of BEC incubated for 48 hours. Conflu exposed to 5-FU at 5 or 10 ug/ml. The total cellul measured after 48 hours incubation. Each point is seven separate experiments, and data in each experim wells.

735

Cell death At 48 hours incubation with 5-FU concentrations of 5 (ig/ml and 10 ng/ml, or MTX at 20 ng/inl and 40 ng/ ml, no effects on cell death were seen in either the BEC ortheHEC. Discussion 5-FU was introduced in 1957 and was for many years used mostly with palliative intent and therefore mainly in low doses, usually administrated as a bolus injection in order to minimize side effects. During the last two decades however, new chemotherapy regimes with curative intent have been introduced, in which 5-FU, administrated as an infusion rather than a bolus, plays an important role. The dosages of 5-FU included in curative treatments are much higher than in palhative ones, which results in increased toxicity and a changed toxicity profile. The pathophysiology of 5-FU induced cardiotoxicity, a side effect known about since the early seventies, still remains unclear, and it is challenging to investigate new ways of explaining its mechanism. One of the most common clinical manifestations of 5-FU induced cardiotoxicity is myocardial ischemia in different guises, a fact which made us focus our interest on the coronary arteries. Previous studies on cultured

In the following discussion we regard the BEC to behave as benign cells with, at least partly, preserved growth control mechanisms, while the HEC were established from a cell line and as such are immortalized

Per cent

Per cent •

p -0.018

00 -

uu -

80 -

80 -

60 -

60 -

p = 0.027

40 -

40 -

20 -

20 -

n -

n Control

20 ng/ml MTX

40 ng/ml MTX

Control

20 ng/ml MTX

40 ng/ml MTX

Figure 6A. Effect of MTX on the release of prostacyclin by BEC, Figure 6B. Effect of MTX on the release of prostacyclin by BEC, expressed as the absolute 6 keto PGFla/100 \i\ medium. BEC expressed as the absolute 6 keto PGFla/total protein. BEC were were incubated with MTX at 20 or 40 ng/ml for 48 hours. The data incubated with MTX at 20 or 40 ng/ml for 48 hours. The experirepresent percentage of control values. Three wells were used for ments were performed seven times and in three wells in each experieach of the seven experiments performed. ment.

Downloaded from http://annonc.oxfordjournals.org/ by guest on June 6, 2016

cells, comparing the toxicity of 5-FU on myocytes and endothelial cells from the rat heart, suggested that myocytes are less susceptible than endothelial cells to the direct toxic effect of 5-FU [13]. This is in agreement with previous studies by our group using other toxic stimuli [20]. Vascular endothelium, previously believed to constitute a selective barrier between blood and tissues, is nowadays regarded as a highly active metabolic and endocrine organ [9] which is involved in a variety of pathophysiological events. Having in mind the importance of vascular endothelium in the regulation of vascular tone and as a regulatory mechanism of thrombogenicity, it seems a logical approach to study this organ under the influence of 5-FU. In an experimental study in rabbits [8], evaluating the immediate effect of 5-FU on vascular endothelium, severe damage to the ultima, occasionally followed by thrombus formation was seen. In further studies [26], assessing the late effects of 5-FU on this structure, 1,3,7,14 and 30 days after treatment, more severe endothelial injury with an increased expression of thrombus formations was found. 5-FLPs effect was most pronounced about 3 days after commencing treatment This corresponds well with clinical observations of 5-FU induced cardiotoxicity, where symptoms usually set in about 3-4 days after the start of treatment. Findings from the studies mentioned above, encouraged us to study endothelial cells in culture, and assess the effect of endotheliumderived substances and how they might be involved in the thrombogenicity of 5-FU observed in animal experiments.

mg, (SD 196). Data represent percentage of their own control values. The release of prostacyclin by BEC remained unchanged with increasing concentrations of MTX when expressed per dish (Figure 6A) or as a fraction of total protein (Figure 6B).

736 with 5-FU and the absence of the same effect in incubations with MTX, could be interpreted as leakage secondary to EC injury and thereby supports a specific susceptibility of EC to 5-FU. Observations of prostacyclin released by BEC incubated for 48 hours correspond well with the results from experimental studies in rabbits [8, 26], where the effect of 5-FU on endothelium in small arteries was evaluated by scanning (SEM) and transmission (TEM) electron microscopy after in vivo treatment with 5-FU. The studies showed that intima injury followed by thrombus formation was more severe 3-7 days after treatment than minutes/ hours after treatment with 5-FU. The observations from the present study, showing an increased release of prostacyclin after 48 hours' incubation with 5-FU, could help to explain findings from the animal studies. These findings taken together could be interpreted to mean that the primary reaction to endothelial injury is an increased release and leakage of vasodilatory, anti-coagulant substances, and only, when this mechanism becomes exhausted do the procoagulant effects take over, being expressed as thrombus formation. Our results were obtained in cell culture and therefore any extrapolation of these results to the in vivo situation should be made with the greatest caution. Being aware of these difficulties, we can, however, conclude that our study gives the support to the hypothesis that 5-FU may cause non-lethal injury to benign endothelium, an unspecific toxicity, which seems to be different from the antitumour effect of 5-FU. Such injury with the possible accompaniment of trombogenesis could be one of the pathophysiologjcal mechanisms behind 5-FU induced cardiotoxicity. Another interesting aspect of 5-FU's injury to endothelium is its possible effect on the endothelium in the vascular bed of the tumor. The increased efficacy of 5-FU, when given by infusion, could be, at least partly, due to the suppressing effect of the drug on tumor vascularization. These aspects deserve great attention and will be investigated in the further studies.

Acknowledgments

The technical assistance of Anna Berglund and Eva Henriksson is gratefully acknowledged.

References 1. Gradishar WJ, Vokes EE. 5-FluorouraciI cardiotoxicity: A critical review. Ann Oncol 1990; 1:409-14. 2. Ensley JF, Patel B, Kloner R et al. The clinical syndrome of 5fluorouracil cardiotoxicity. Invest New Drugs 1989; 7:101-9. 3. Robben N, Pippas A, Moore J. The syndrome of 5-fluorouracil cardiotoxicity. Cancer 1993; 71(2): 493-509. 4. Freeman N, Constanza M. 5-Fluorouracil-associated cardiotoxicity. Cancer 1988; 61: 36-45. 5. Burger AJ, Mannino S. 5-Fluorouracil induced coronary vasospasm. Am Heart J 1989; 114:433-6.

Downloaded from http://annonc.oxfordjournals.org/ by guest on June 6, 2016

and have 'half-malignant1 growth characteristics. The study showed no effect of either 5-FU or MTX on cell death in either the BEC or the HEC, i.e., no cell loss or membrane damage was detected. Instead sub-lethal injury, manifested as changes in (3H)thymidine incorporation and release of prostacyclin, was registered. Our results thus support the assumption that the observed changes in DNA synthesis and prostacyclin release were not due to cell loss but instead may reflect true changes in these functions, probably as a consequence of sub-lethal toxicity. To further support this we performed some experiments where incubation was continued after 48 h but the medium was changed to a medium without drugs. Here the cultures behaved quite similarly to each other, i.e., the effects of the drugs were completely reversible. This effect was probably different from the cytotoxic effects of the chemotherapeutic drugs studied. The DNA synthesis, measured as (3H)thymidine incorporation, decreased significantly in both HEC and BEC when incubated with 5-FU for 48 hours. This effect was not seen in the same experiments with MTX, which indicates a specific susceptibility of EC to 5-FU. We do not believe that the observed differences were due to incomparability of the selected concentrations of the drugs, because the concentrations were carefully selected on the basis of similar cell culture experiments and clinical practice. The selection of a MTX concentration used in the present study was based on the in vitro experiment evaluating the effect of low-dose MTX on DNA synthesis by human umbilical vein EC [14]. In this study, a significant inhibition of (3H) deoxyuridine incorporation into EC was observed at a MTX concentration of 5 x 10~9 M, which is attained in the serum of patients treated for rheumatoid arthritis [16, 17] and maximal inhibition was observed at a concentration of 10"7 M. The mean MTX concentration value was then worked out and found to be about 20 ng/ml, which correspond to the conventional dose of the drug, up to 100 mg/m2, used in the clinical oncological practice. The dose of MTX required to produce toxicity varies between different organs, a critical concentration threshold must be exceeded before organ toxicity will occur. For bone marrow and gastrointestinal epithelium the plasma concentration and time threshold appear to be 2 x 10"8 M and about 42 hours respectively [15]. The concentrations of MTX used in our study exceeded this threshold, and therefore we found them adequate to study the effect of MTX on EC. Possibly, at higher concentrations of MTX, we might seen more substantial effects of MTX on EC, however, the finding of the increasing DNA synthesis in BEC, during incubation with increasing concentrations of MTX, is rather against such expectation. The decrease in (3H)thymidine incorporation was greater in BEC than in HEC which could be interpreted to mean that benign endothelial cells are more sensitive to 5-FU than half-malignant ones. The release of prostacyclin by BEC when incubated

737

19. 20.

21. 22.

23.

24. 25. 26.

patients on a protracted venous infusion with or without interferone. Ann Oncol 1996; 7:47-53. Witte L, Kaplan K, Nossel H et al. Studies of the release from human platelets of the growth factor for cultured human arterial smooth muscle cells. Circ Res 1978; 42:402-9. Stavenow L, Falke P, Berglund A. Effects of different injurious stimuli on cell death, proliferation, and collagen secretion by rabbit aortic smooth muscle cells and human umbilical vein endothelial cells in culture. Med Biol 1983; 61: 214-8. Lowry OH, Rosenbourg NJ, Farr AL et al. Protein measurements with folin phenol reagent J Biol Chem 1951; 193: 26575. Falke P, Mattiasson I, Stavenow L et al. Effects of a competitive inhibitor (mevinolin) of 3-hydroxy-3-methylglutaryl coenzym A reductase on human and bovine endothelial cells, fibroblasts and smooth muscle cells in vitro. Pharmacol & Toxicol 1989; 64:173-6. Pessah-Rasmussen H, Stavenow L, Seidegird J-E et al. Human fibroblasts lacking trans-stilbene oxide active glutathione transferase exhibit increased cell death when exposed to polycyclic aromatic hydrocarbons. Pharmacol & Toxicol 1992; 70:361-5. Eckel RW, Fujimoto WY. Quantification of cell death in human fibroblasts by measuring the loss of 14C-thymidine from prelabelled cell monolayers. Anal Biochem 1981; 114:118-24. Edgell CIS, McDonald CC, Graham JB. Permanent cell line expressing factor VIII related antigen established by hybridization. Proc Natl Acad Sci USA 1983; 80: 3734-7. Cwikiel M, Eskilsson J, Wieslander JB, Stjernquist U, Albertsson M. The appearance of endothelium in small arteries after the treatment with 5-fluorouracil. An electron microscopic study of late effects in rabbits. Scanning Microsc 1996; 10(3): in press.

Received 22 February 1996; accepted 30 July 1996. Correspondence to: Magdalena Cwikiel, MD Department of Oncology University Hospital 221 85 Lund Sweden

Downloaded from http://annonc.oxfordjournals.org/ by guest on June 6, 2016

6. Stevenson DL, Mikhailidis DP, Gillett DS. Cardiotoxicity of 5-fluorouracil. Lancet 1977; 2: 406-7 (Letter). 7. Kuzel T, Esyaraz B, Green D et al. Thrombogenicity of intravenous 5-fluorouracil alone or in combination with cisplatin. Cancer 1990; 65: 885-9. 8. Cwikiel M, Albertsson M, Zhang B et al. The influence of 5-fluorouracil on the endothelium in small arteries. A scanning and transmission electron microscopic study in rabbits. Scan Micros 1995; 9(2): 561-76. 9. Vane JR, Anggard EE, Botting RM. Regulatory functions of the vascular endothelium. N Engl J Med 1990; 323: 27-36. 10. Nimmrich H, Meaires S, Forssman et al. Scanning electron microscopy of endothelium. In Didio LJT, Motta PM, Allen DJ (eds): Three Dimensional Microanatomy of Cells and Tissue Surfaces. Amsterdam: Elsevier 1981. 11. Mason R, Sharp D, Chuang H. The endothelium. Role in thrombosis and haemostasis. Arch Pathol Lab Med 1977; 101: 61. 12. Xu CB, Stavenow L, Pessah-Rasmussen H. Interactions between cultured bovine arterial endothelial and smooth muscle cells; studies on the release of prostacycline by endothelial cells. Artery 1992; 19(2): 94-111. 13. Wenzel DG, Cosma GN. A model system for measuring comparative toxicities of cardiotoxic drugs for cultured rat heart myocytes, endothelial cells and fibroblasts. II. Doxorubicin, 5-fluorouracil and cyclophosphamide. Toxicol 1984; 33: 117— 28. 14. Hirata S, Matsubara T, Saura R et al. Inhibition of in vitro vascular endothelial cell proliferation and in vivo neovascularisation by low-dose methotrexate. Arthritis Rheum 1989; 32(2): 1065-73. 15. Schornagel JH, Me Vie. The clinical pharmacology of methotrexate. Cancer Treat Rev 1983; 10: 53-75. 16. Edelman J, Biggs DF, Jamali F et al. Low-dose methotrexate kinetics in arthritis. Clin Pharmacol Ther 1984; 35: 382-6. 17. Kremer JM, Galivan J, Streckfuss A et al. Methotrexate metabolism analysis in blood and liver of rheumatoid arthritis patients: Association with hepatic folate deficiency and formation of polyglutamates. Arthritis Rheum 1986; 29: 832-5. 18. Findlay MPN, Raynaud F, Cunningham D et al. Measurement of plasma 5-fluorouracil by high-performance liquid chromatography with comparison of results to tissue drug level observed using in vivo 19F magnetic resonance spectroscopy in