Vascular endothelial growth factor in peritoneal dialysis: A longitudinal follow-up

Vascular endothelial growth factor in peritoneal dialysis: A longitudinal follow-up MACHTELD M. ZWEERS, DIRK G. STRUIJK, WATSKE SMIT, and RAYMOND T. KREDIET AMSTERDAM, THE NETHERLANDS

In a previous study, vascular endothelial growth factor (VEGF) was found to be locally produced in the peritoneal tissue of patients undergoing peritoneal dialysis (PD) who were being treated with glucose-containing PD solutions. Locally produced VEGF (LVEGF) was positively related to the mass transfer area coefficient (MTAC) of creatinine and to glucose absorption, both of which are representative of the peritoneal vascular surface area. It was therefore hypothesized that VEGF is involved in the peritoneal neoangiogenesis found in long-term PD. The aim of the present study was to investigate the time course of peritoneal VEGF levels in PD patients treated with glucose-based PD solutions during longitudinal follow-up. We also studied the effect of the switch to glucose-free PD treatment on VEGF production. Forty standard peritoneal permeability analyses (SPAs) with 3.86% glucosecontaining dialysis solution were investigated. The SPAs were performed in 10 PD patients with a median number of three SPAs per patient during a follow-up of 23 months. Duration of PD treatment at the last SPA was 74 months. All patients were initially treated with glucose-containing dialysis solutions. Four patients switched after 114 months of glucose-based PD to glucose-free PD and were followed for 7 months. A PD regimen of icodextrin, glycerol, and amino acid–based dialysis solutions was applied in these patients. Four SPAs were performed per patient in this period. To predict the VEGF dialysate-to-serum ratio (D/S), when diffusion would be the only explanation for the VEGF dialysate concentration, we calculated the power relationship between D/S ratios of serum proteins that are only transported across the peritoneum and the molecular weights of those proteins. The measured VEGF D/S ratio was higher than expected (P < .001) in each observation, pointing to local production of VEGF. LVEGF increased with duration of glucose PD, 11.7 ng/L to 23.45 ng/L (P < .03). LVEGF decreased in all 4 patients undergoing glucose-free PD, from 57.35 ng/L to 23.10 ng/L. A correlation (r = 0.83, P < .001) was found between the differences in MTAC creatinine between the first and last SPA during glucose-based PD and the difference in LVEGF between these observations. A similar correlation was present between the difference in glucose absorption and the difference in LVEGF (r = 0.85, P < .001). This supports a pathogenetic role of high glucose dialysate concentrations in the development of changes in the peritoneum that are found in long-term PD. Treatment with non-glucose-based PD solutions may inhibit further development of these alterations. (J Lab Clin Med 2001;137:125-32)

Abbreviations: D/S ratio = dialysate-to-serum ratio; EDTA = ethylenediaminetetraacetic acid; IgG = immunoglobulin G; LVEGF = peritoneal effluent concentration attributed to local production of VEGF; MTAC = mass transfer area coefficient; PD = peritoneal dialysis; SPA = standard peritoneal permeability analysis; VEGF = vascular endothelial growth factor

From the Department of Nephrology, Academic Medical Center Amsterdam, and the Dianet Foundation Utrecht-Amsterdam. Supported by Grant C95-5009 from the Dutch Kidney Foundation.

Reprint requests: M. M. Zweers, MSc, Department of Nephrology, F4.215, PO Box 22700, 1100 ED Amsterdam, The Netherlands. Copyright © 2001 by Mosby, Inc.

Submitted for publication July 5, 2000; revision submitted October 9, 2000; accepted October 13, 2000.

0022-2143/2001 $35.00 + 0 5/1/112235 doi:10.1067/mlc.2001.112235 125

126 Zweers et al

Vascular endothelial growth factor is a mediator in the development of neoangiogenesis in proliferative diabetic retinopathy.1-5 Local production in the peritoneal cavity has recently been reported by us6 in patients undergoing long-term PD with glucose-containing dialysis solutions. The results of this study suggested a role for VEGF in the development of peritoneal neoangiogenesis, as found in long-term PD patients.7 Other investigators8-10 described an increase in the D/P ratios or MTACs of low-molecular-weight solutes with the duration of peritoneal dialysis. These functional parameters are likely to reflect the peritoneal vascular surface area.11,12 We found relationships between the effluent concentration attributed to LVEGF and the MTAC of creatinine and also with glucose absorption from the peritoneal cavity. Furthermore, inverse correlations were seen between the LVEGF and the transcapillary ultrafiltration and the net ultrafiltration. These observations all suggest a causal relationship between prolonged peritoneal dialysis with glucose-based dialysis solutions and the development of a large peritoneal vascular surface area in association with, or mediated by, peritoneal production of VEGF. Therefore the aim of the present study was to investigate the presence of VEGF in the peritoneal effluent of patients treated with glucose-containing peritoneal dialysis solutions in a longitudinal follow-up in relation to their peritoneal transport parameters. The second purpose was to study the effect of a switch to glucose-free dialysis treatment on LVEGF and peritoneal permeability characteristics in long-term patients with ultrafiltration failure. METHODS Patients. Ten non-diabetic patients undergoing PD (5 men and 5 women) were investigated with SPAs. Forty SPAs were performed, with a median number of 3 per patient (range 2 to 6) during a median follow-up of 23 months (range 7 to 37). The median duration of PD at the time of the last investigation was 74 months (range 16 to 152). All patients were initially treated with commercially available glucose-based dialysis solutions (Dianeal; Baxter, Utrecht, The Netherlands). Four patients switched after 114 months (range 80 to 149) of glucose PD to glucose-free dialysis treatment because of ultrafiltration failure. This was defined as net ultrafiltration less than 400 mL during a 3.86% glucose 4-hour dwell.13 The glucose-free dialysis regimen was composed of a combination of a 2.5% glycerol-based dialysis solution that is not commercially available; commercially available Nutrineal, a 1.1% amino acid–containing dialysis solution; and Extraneal, which contains 7.5% icodextrin as an osmotic agent (Baxter, Utrecht, The Netherlands). During a follow-up of 7 months (range 2 to 13), these patients underwent a median of 4 SPAs per patient (range 2 to 5). None of the patients had peritonitis at the time of the study or during the 4 weeks preceding the investigation. Informed consent was obtained from all

J Lab Clin Med February 2001

patients after an explanation of the purpose and the methods of the study. Study design. Previously described standard peritoneal permeability analyses14 were performed with 3.86% glucosecontaining dialysate (Dianeal). In brief, before the instillation of the test solution, the peritoneal cavity was rinsed with 1.36% glucose dialysate. The volume marker dextran 70, 1 g/L (Hyskon; Medisan Pharmaceuticals, Uppsala, Sweden), was added to the test solution for the calculation of fluid kinetics.15 Samples were taken from the test solution before inflow and at 10, 20, 30, 60, 120, 180, and 240 minutes after instillation of the test solution. These samples were obtained after temporal drainage of approximately 200 mL to avoid a dead space effect. Subsequent to the total drainage at 240 minutes, the peritoneal cavity was rinsed again with 1.36% Dianeal. A sample from this rinse was used to calculate the residual volume. Blood was drawn at the start and at the end of the test. After the first blood sample was obtained, dextran 1 (20 mL; Promiten; NBPI, Emmer-Compascuum, The Netherlands) was given intravenously to prevent a possible anaphylaxis to dextran 70.16 Assays. Glucose concentrations in plasma and dialysate samples were assessed by the glucose oxidase–peroxidase method with an autoanalyzer (SMA-II; Technicon, Tarrytown, NY). Urea and creatinine were measured with enzymatic methods by an automated analyzer (Hitachi H747; Boehringer Mannheim, Mannheim, Germany). Total protein in plasma was determined by biuret methods (Roche, Almere, The Netherlands) and also with an automated analyzer (Hitachi 747; Boehringer Mannheim). β2-Microglobulin was determined with an Imx system applying a microparticle enzyme immunoassay (Abbott Laboratories, North Chicago, IL). Albumin, IgG, and α2-macroglobulin were measured by nephelometry (BN 100; Behring, Marburg, Germany). VEGF was measured in dialysate and serum as previously described.6 In brief, a commercially available enzyme-linked immonosorbent assay (human VEGF, Quantikine; R&D Systems, Minneapolis, MN) was applied, with a lower detection limit of 15.0 ng/L. Serum was allowed to clot for 4.5 hours. Before determination of the growth factor in dialysate, the effluent samples were concentrated 9.5 times (range 7.6 to 12.4) by positive pressure ultrafiltration with a 250-mL cell and a YM-10 membrane with a molecular cutoff point of 10 kd (Amicon Corp, Danvers, MA). The concentration factor was defined as the quotient of the albumin concentration in the concentrate divided by the albumin concentration in the original effluent sample. Total dextran 70 was determined by high-performance liquid chromatography.17 Calculations. Peritoneal fluid and solute kinetics were calculated as reported previously.14,15 In brief, fluid transport across the peritoneal membrane is influenced by opposing mechanisms. The transcapillary ultrafiltration increases the intraperitoneal volume, and fluid loss from the peritoneal cavity is assumed to occur through transcapillary back filtration and uptake into the lymphatic system. The result of these is the net ultrafiltration. The transcapillary ultrafiltration was calculated from the dilution of the volume marker. The convective disappearance of the volume marker from the peritoneal cavity can be used as an indirect measure to quantify

J Lab Clin Med Volume 137, Number 2

the contribution of the peritoneal lymphatics in the absorption of fluid from the peritoneal cavity.18 These calculations of the effective lymphatic absorption include all pathways of uptake into the lymphatic system, both interstitial and subdiaphragmatic. The change in intraperitoneal volume during the dwell can be assessed from the dilution of dextran 70 after correction for incomplete recovery. The net ultrafiltration is the difference between the transcapillary ultrafiltration and the effective lymphatic absorption. To express the transport of the low-molecular-weight solutes urea and creatinine, MTACs were calculated according to the model of Waniewski et al.19 The solute concentrations in plasma were corrected for plasma water. 20 Glucose absorption was estimated as the difference between the instilled and the recovered amount of glucose, relative to the instilled quantity of glucose in the dialysate. The peritoneal handling of the macromolecules β2-microglobulin, VEGF, albumin, IgG, and α2-macroglobulin was expressed as D/S ratio. A peritoneal transport line was computed for all investigations performed in each patient, based on the least squares regression analysis of the D/S ratio of β2-microglobulin (mol wt 11.8 kd), albumin (mol wt 69 kd), IgG (mol wt 150 kd), and α2-macroglobulin (mol wt 820 kd) and their molecular weights when plotted on a double logarithmic scale.21 The slope of this line represents the size selectivity of the peritoneal membrane as these proteins are transported from the circulation to the peritoneal cavity. By interpolation of the molecular weight of VEGF (mol wt = 34 kd) in the regression equation, the expected D/S ratio was calculated, assuming the dialysate concentration of VEGF would be determined only by transport from the circulation. The concentration of the growth factor attributed to local production was defined as the difference between the measured and expected dialysate concentration. Statistics. The results are presented as median values and ranges. The differences between the measured and expected dialysate concentrations of VEGF were tested by a modified t test to determine whether the deviation from the regression line was significant. This test takes the variability of the regression lines into account.22 To investigate the absolute difference between the measured and expected dialysate concentrations of VEGF, the Wilcoxon matched pairs rank sum test was applied. The differences in the effluent concentrations attributed to local production in the longitudinal follow-up of glucose-based PD were investigated with repeated measurement analysis of variance. Spearman rank correlation analysis was used for the calculation of correlation coefficients. RESULTS

Parameters of peritoneal permeability obtained from the last SPA during glucose PD treatment are presented in Table I. Regression lines were calculated for each individual patient for each investigation on the basis of the power relationship between the D/S ratio of the various serum proteins and their molecular weights (see Methods). The r values of the regression lines all exceeded 0.91 (P < .05). Based on these regression

Zweers et al

127

Table I. Parameters of peritoneal permeability Glucose PD (n = 10)

Fluid transport TCUF (mL/4 h) ELA (mL/4 h) NUF (mL/4 h)

667 (185-1059) 250 (89-523) 579 (81-1126)

Non-glucose PD (n = 4)

225 (126-796) 132 (108-194) 41 (–16-602)

Solute transport MTAC urea (mL/min) 17.7 (13.4-20.9) 15.5 (8.5-19.1) MTAC creatinine (mL/min) 9.0 (7.0-18.0) 10.6 (6.6-16.6) Glucose absorption (%) 60 (44-75) 64 (44-76) Effluent VEGF attributed 17.3 (6.5-225.3) 23.1 (4.4-61.6) to local production (ng/L) Values are presented as median and range for data obtained from the last SPA during glucose PD and the last glucose-free investigation. TCUF, Transcapillary ultrafiltration; ELA, effective lymphatic absorption, NUF, net ultrafiltration; MTAC, mass transfer area coefficient.

lines, the D/S ratio of VEGF could be predicted when the concentration in the effluent would only be determined by transport from the circulation to the peritoneal cavity (Fig 1). The measured D/S ratio of VEGF was significantly greater (P < .001) than could be expected based on the peritoneal transport line, indicating local production of VEGF (LVEGF). The follow-up of glucose PD in 9 patients in relation to LVEGF is plotted in Fig 2, A. Three patients had a total of 5 peritonitis episodes; the time points of these are indicated in Fig 2. Nine patients are presented because 1 patient underwent only 1 SPA during glucose dialysis treatment before switching to the glucose-free treatment. LVEGF increased with prolonged treatment of glucose PD, from 11.7 ng/L (2.0 to 19.0) to 23.5 ng/L (5.4 to 75.9), P = .06. A comparison between the first and last LVEGF showed a significant rise (P < .03, Table II). The magnitude of the increase in LVEGF was independent from the initial LVEGF and independent from the duration of PD at the time of the first investigation. When the LVEGF obtained from the last SPA during glucose-based PD was compared with the LVEGF from the last SPA during glucose-free PD, a decrease was measured (Table II). A correlation (r = 0.83, P < .001) was found between the difference in the MTAC of creatinine in the initial SPA and last SPA during glucose PD and the difference between LVEGF values in the same investigations (Fig 3). A similar correlation (r = 0.85, P < .001) was present between the difference in peritoneal glucose absorption and the difference in LVEGF (Fig 3). The changes in these parameters obtained in the patients after the switch to glucose-free PD are shown in the same figure.

J Lab Clin Med February 2001

128 Zweers et al

Fig 1. Regression line based on the power relationship between the D/S ratio of α2-microglobulin, albumin, IgG, and β2-microglobulin () and their molecular weights. The measured D/S ratio of VEGF () is given in relation to its molecular weight. Values are presented as mean and SD. *P < .001.

Table II. Follow-up of VEGF attributed to local production in peritoneal effluent

Glucose PD Non-glucose PD

Initial glucose PD LVEGF (ng/L)

Last glucose PD LVEGF (ng/L)

Last non-glucose LVEGF (ng/L)

Months of follow-up

n

P

11.7 (2.0-19.0) —

23.5 (5.4-75.9) 57.4 (8.2-225.3)

— 23.1 (4.4-61.7)

24 (2-37) 7 (2-13)

9 4

.03 NS

Median and ranges are presented.

DISCUSSION

In this study the presence of VEGF in peritoneal effluent of patients treated with glucose PD was investigated during longitudinal follow-up in relation to peritoneal permeability parameters. Furthermore, the effect of the switch to glucose-free dialysis treatment on VEGF production and the characteristics of peritoneal permeability was studied in patients with ultrafiltration failure. The concentration of VEGF in peritoneal effluent found in each patient in each observation was higher than could be explained by diffusion from the circulation. This points to local peritoneal production of VEGF as was suggested by us previously.6 In the present study, VEGF was measured in serum. Recently, platelet-mediated secretion of VEGF during clotting processes has been described by Webb et al.23 As a consequence, serum levels are on average twice the concentration measured in EDTA-treated plasma24-26 or citrated plasma.27 We used a standardized clotting time of 4.5 hours before centrifugation. If VEGF would have

been measured in EDTA-treated plasma, the dialysateto-plasma ratio would have been higher. Therefore the calculated local production of VEGF in the present study may have been an underestimation of the actual LVEGF. However, VEGF can exist in any of five alternatively spliced mRNA forms.3 Two common species give rise to protein species of VEGF165 (34 kd) and VEGF121 (25 kd), which are soluble and could have been present in plasma or serum and in dialysis effluent. Currently no immunoassay is available that differentiates between these two VEGF isoforms. The possibility exists that the VEGF measured in our analysis is of a lower molecular weight form than the assumed 34 kd. When the expected D/S ratios were calculated on the basis of a VEGF molecular weight of 25 kd (0.031, range 0.010 to 0.106), the measured VEGF D/S ratios (0.093, range 0.018 to 0.826) were still significantly different from the expected D/S ratio when taking the variability of the transport line into account22 (P < .01), indicating local production. The production of VEGF attributed to local produc-

J Lab Clin Med Volume 137, Number 2

Zweers et al

129

A

B Fig 2. Individual data of the longitudinal follow-up of VEGF attributed to local production in the peritoneum obtained in 9 patients during glucose-based PD are presented in B. These data obtained in 4 long-term PD patients before and after the switch to glucose-free PD treatment are presented in A. Time point t = 0 is the last observation during glucose-based PD; thereafter the patients were treated with glucose-free dialysis solutions. Three patients experienced a total of 5 peritonitis episodes (arrows).

tion increased with prolonged duration of glucose PD. LVEGF decreased in all four patients with ultrafiltration failure after the switch to glucose-free dialysis treatment. This could imply that the high glucose concentrations in dialysis solutions may be a mediating factor in the triggering of VEGF production and the development of peritoneal neoangiogenesis,7 in analo-

gy with VEGF production induced by hyperglycemia in diabetic retinopathy.1,2,5 Low-molecular-weight solutes such as creatinine and glucose have been shown not to be size-selectively hindered by the peritoneal membrane11,12; therefore their transport to the peritoneal cavity mainly depends on the peritoneal vascular surface area. Consequently, changes

J Lab Clin Med February 2001

130 Zweers et al

A

B

Fig 3. Difference between the first and last MTAC creatinine and the difference between the first and last observation of LVEGF during glucose-based PD () and the last LVEGF determination during glucose-based PD and the last one during glucose-free PD (■) are shown (A, n = 10, r = 0.83, P < .001). A similar relationship was present between the difference in glucose absorption and the difference in LVEGF (B, n = 10, r = 0.85, P < .001).

in MTACs or glucose absorption reflect changes in the peritoneal vascular surface area. In our cross-sectional study,6 MTAC creatinine and the glucose absorption correlated with LVEGF, suggesting involvement of VEGF in peritoneal neoangiogenesis. The relationship that was present between the difference in MTAC creatinine between the first and the last observation on glucose PD and the difference in LVEGF between these investigations emphasizes a possible role for VEGF

mediating the peritoneal vascular surface area during glucose PD, especially because the difference in these parameters obtained in the last SPA during glucose PD and those obtained in the last glucose-free investigation showed the same relationship: LVEGF decreased after the switch to glucose-free dialysis treatment, and the parameters of the peritoneal vascular surface area also decreased in 3 out of 4 patients. Significant correlations were still present when the glucose-free group

J Lab Clin Med Volume 137, Number 2

was left out. Spearman rank correlation coefficients for the relationship between ∆MTACcreatinine and ∆LVEGF in the 9 glucose patients was r = 0.88, P < .002, and r = 0.63, P < .07 for the relationship between ∆glucose absorption and ∆LVEGF. These findings support a possible relationship between the peritoneal exposure to extremely high glucose concentrations and the development of peritoneal angiogenesis through a pathogenetic mechanism with the involvement of VEGF. The magnitude of the increase in LVEGF during glucose PD was, however, independent from the initial LVEGF and the duration of PD at the start of the follow-up during glucose-based PD. Some patients showed a steep rise in LVEG during less than 3 years of follow-up of glucose-based PD, whereas in others, LVEGF increased only marginally during the same period of follow-up. This could explain why no relationship between LVEGF and the duration of PD was found in our cross-sectional study.6 The great interindividual variability suggests a role for individually determined factors that may influence the effect of the prolonged duration of glucose PD on the slope of, and the extent of, the local production of VEGF in each PD patient. These factors may involve VEGF and other polymorphisms. The steep increase in LVEGF in some patients may also have been mediated by the development of impaired ultrafiltration capacity, leading to an increased use of a 3.86% glucose-based dialysis solution. This may have been a potent stimulation of VEGF production. These speculations, however, require further study. We thank Natalie Schouten, Nicole van den Berg, Laura Splint, and Johan Hirallal for technical assistance. REFERENCES

1. Aiello LP, Avery RL, Arrigg PG, Keyt BA, Jampel HD, Shah ST, et al. Vascular endothelial growth factor in ocular fluid of patients with diabetic retinopathy and other retinal disorders. N Engl J Med 1994;331:1480-7. 2. Tanaka Y, Katoh S, Hori S, Miura M, Yamashita H. Vascular endothelial growth factor in diabetic retinopathy. Lancet 1997;349:1520. 3. Dvorak HF, Brown LF, Detmar M, Dvorak AM. Vascular permeability factor/vascular endothelial growth factor, microvascular hyperpermeability, and angiogenesis. Am J Pathol 1995;146:1029-39. 4. Brown LF, Detmar M, Claffey K, Nagy JA, Feng D, Dvorak AM, et al. Vascular permeability factor/vascular endothelial growth factor: a multifunctional angiogenic cytokine. In: Goldberg ID, Rosen EM, editors. Regulation of angiogenesis. Basel: Birkhäuser Verlag; 1997. p. 233-69. 5. Ferrara N. Vascular endothelial growth factor. The trigger for neovascularization in the eye. Lab Invest 1995;72:615-8. 6. Zweers MM, de Waart DR, Smit W, Struijk DG, Krediet RT. Growth factors VEGF and TGF-β1 in peritoneal dilaysis. J Lab Clin Med 1999;134:124-32.

Zweers et al

131

7. Mateijsen MAM, Van der Wal AC, Hendriks PMEM, Zweers MM, Mulder J, Struijk DG, Krediet RT. Vascular and interstitial changes in the peritoneum of CAPD patients with peritoneal sclerosis. Perit Dial Int 1999;19:517-25. 8. Davies SJ, Bryan J, Phillips L, Russell GI. Longitudinal changes in peritoneal kinetics: effects of peritoneal dialysis and peritonitis. Nephrol Dial Transplant 1996;11:498-506. 9. Selgas R, Fernadez-Reyes M-J, Bosque E, Bajo M-A, Borrego F, Jimenez C, et al. Functional longevity of the human peritoneum: how long is continuous peritoneal dialysis possible? Results of a prospective median long-term study. Am J Kidney Dis 1994;23:64-73. 10. Krediet RT, Zemel D, Imholz ALT, Koomen GCM, Struijk DG, Arisz L. Indices of peritoneal permeability and surface area. Perit Dial Int 1993;13:S31-4. 11. Struijk DG, Krediet RT, Koomen GCM, Hoek FJ, Boeschoten EW, von Reijden HJ, et al. Functional characteristics of the peritoneal membrane in long-term continuous ambulatory peritoneal dialysis. Nephron 1991;59: 213-20. 12. Krediet RT, Zemel D, Imholz ALT, Struijk DG. Impact of surface area and permeability on solute clearances. Perit Dial Int 1994;14:(suppl 3):S70-7. 13. Ho-dac-Pannekeet MM, Atasever B, Struijk DG, Krediet RT. Analysis of ultrafiltration failure in peritoneal dialysis patients by means of standard peritoneal permeability analysis. Perit Dial Int 1997;17:144-50. 14. Pannekeet MM, Imholz ALT, Struijk DG, Koomen GCM, Langedijk MJ, Schouten N, et al. The standard peritoneal permeability analysis: a tool for the assessment of peritoneal permeability characteristics in CAPD patients. Kidney Int 1995;48:866-75. 15. Krediet RT, Struijk DG, Koomen GCM, Arisz L. Peritoneal fluid kinetics during CAPD measured with intraperitoneal dextran 70. ASIO Trans 1991;37:662-7. 16. Renck H, Ljungström HG, Hedin H, Richter W. Prevention of dextran induced anaphylactic reaction by hapten inhibition. Acta Chir Scan 1983;149:355-60. 17. Koomen GCM, Krediet RT, Leegwater ACJ, Struijk DG, Arisz L, Hoek FJ. A fast reliable method for the measurement of intra-peritoneal dextran 70, used to calculate lymphatic absorption. Adv Perit Dial 1991;7:10-4. 18. Krediet RT. Fluid absorption in the peritoneum: it is less simple than you thought. Nephrol Dial Transplant 1994;9:341-3. 19. Waniewski J, Werynski A, Heimbürger O, Lindholm B. Simple models for description of small-solute transport in peritoneal dialysis. Blood Purif 1991;9:129-41. 20. Waniewski J, Heimbürger O, Werynski A, Lindholm B. Aqueous solute concentrations and evaluation of mass transport coefficients in peritoneal dialysis. Nephrol Dial Transplant 1992;7:50-6. 21. Zemel D, Imholz ALT, de Waart DR, Dinkla C, Struijk DG, Krediet RT. Appearance of tumor necrosis factor-α and soluble TNF receptors I and II in peritoneal effluent of CAPD. Kidney Int 1994;46:1422-30. 22. Reddingius RE, Schröder CH, Daha MR, Willems JL, Koster AM, Monnens LAH. Complement in serum and dialysate in children on continuous ambulatory peritoneal dialysis. Perit Dial Int 1995;15:49-53. 23. Webb NJA, Bottomley MJ, Watson CJ, Brenchley PEC. Vascular endothelial growth factor (VEGF) is released from platelets during blood clotting: implications for measurement of circulating VEGF levels in clinical disease. Clin Sci 1998;94:395-404. 24. Mohle R, Green D, Moore MA, Nachman RL, Rafii S. Con-

132 Zweers et al

stitutive production and thrombin-induced release of vascular endothelial growth factor by human megakaryocytes and platelets. Proc Natl Acad Sci USA 1997;94:663-8. 25. Maloney JP, Silliman CC, Ambruso DR, Tuder RM, Voelkel NF. In vitro release of vascular endothelial growth factor during platelet aggregation. Am J Physiol 1998;275:H1054-61. 26. Salven P, Orpana A, Joensuu H. Leukocytes and platelets of

J Lab Clin Med February 2001

patients with cancer contain high levels of vascular endothelial growth. Clin Cancer Res 1999;5:487-91. 27. Banks RE, Forbes MA, Kinsey SE, Stanley A, Ingham E, Walters C, et al. Release of the angiogenic cytokine vascular endothelial growth factor (VEGF) from platelets: significance for VEGF measurements and cancer biology. Br J Cancer 1998;77:956-64.