Reactive Oxygen Species Production by Monocytes and Polymorphonuclear Leukocytes During Dialysis

Reactive Oxygen Species Production by Monocytes and Polymorphonuclear Leukocytes During Dialysis Jonathan Himmelfarb, MD, J. Michael Lazarus, MD, and Raymond Hakim, MD, PhD • Intradialytic production of reactive oxygen species (ROS) by monocytes and polymorphonuclear leukocytes (PMNL) was examined separately in six hemodialysis patients. Samples obtained 15 minutes after initiation of dialysis with new cuprophane membanes demonstrated significantly increased (P < 0.05) ROS production in both cell populations as measured by the fluorescence of a specific intracellular marker (2',7'-dichlorofluorescein diacetate [DCFH-DA)) assayed by flow cytometry. Granulocytes harvested during dialysis also showed decreased responsiveness to exogenous C5a and F-Met-Leu-Phe (FMLP) at 15, 30, and 60 minutes after initiation of dialysis (P < 0.05). Our data suggest that hemodialysis with cuprophane membrane is associated with monocyte and PMNL activation as shown by production of ROS coincident with peak activation of the complement cascade; these granulocytes become refractory to further stimulation with C5a and FMLP during dialysis. © 1991 by the National Kidney Foundation, Inc. INDEX WORDS: Monocytes; interleukin; fluorescence; cuprophane; C5a; F-Met-Leu-Phe.

T

HE "INTERLEUKIN HYPOTHESIS" has been suggested as a potential explanation for several dialysis-related adverse signs and symptoms.1.2 This hypothesis assumes that monocyte activation occurs during dialysis, either secondary to complement activation, by dialysate factors, or by direct contact of monocytes with dialysis membranes. 3 To date there has been no direct confirmation of monocyte activation during clinical dialysis. Although products of monocyte activation such as interleukin-l (IL-l) and tumor necrosis factor-a (TNF-a) have been detected in dialysis patients recently, their serum concentration does not appear to correlate with the type of dialysis membrane, and specific evidence for monocyte activation in vivo has not been presented. 4-9 Isolation of monocytes from mononuclear or granulocyte cell fractions is difficult and subject to several technical problems, which are often associated with artifactual increase in activation. However, flow cytometric analysis can differentiate different cell types by size and granularity and allows the separate analysis of monocytes and polymorphonuclear leukocyte (PMNL) fractions. We therefore used this novel technique for the detection of intracellular reactive oxygen species (ROS) as an index of monocyte activation specifically and leukocyte activation in general during dialysis. Intracellular ROS, a measure of granulocyte and monocyte activation, was detected by a method developed by Bass et al lO in which a dye, 2'7'-dichlorofluoroscein diacetate (DCFH-DA), rapidly diffuses across cell membranes and is trapped

within the cell by a deacetylation reaction. In the presence of hydrogen peroxide, this compound is oxidized to 2 '7 '-dichlorofluorescein (DCF), which is highly fluorescent. Once oxidized, DCF remains stable, thus allowing for quantitative measurement of intracellular hydrogen peroxide production. 10 Our studies demonstrate that dialysis with cuprophane membrane is associated with monocyte and PMNL activation that occurs at the time of maximum complement activation, namely 15 minutes. 11 This activation further results in transient refractoriness of the PMNL to further stimulation by complement products and F-Met-Leu-Phe. MATERIALS AND METHODS

Patient Characteristics Six patients on chronic maintenance hemodialysis were selected for study. Informed consent was obtained from the subjects and the project was approved by the Institutional Review Board of the Hospital. The mean age of the patients was 57.5 years (range, 35 to 73), with a mean time on dialysis of 55

From the Department of Medicine, University of ~rmont Medical School, Maine Medical Center, Portland, ME, Harvard Medical School, and the End-Stage Renal Disease Program, Brigham & Women's Hospital, Boston, MA; and the Department of Medicine, Division of Nephrology, Vanderbilt University Medical Center, Nashville, TN. Supported in part by Grant No. ROI-HL36015 from the National Institutes of Health. Address reprint requests to Raymond M. Hakim, MD,PhD, Division of Nephrology, Vanderbilt University, B-2214 Medical Center North, Nashville, TN 37232-2373. © 1991 by the National Kidney Foundation, Inc. 0272-6386/91/1703-0005$3.00/0

American Journal of Kidney Diseases, Vol XVII, No 3 (March). 1991: pp 271-276

271

272 months (range, 15 to 156). Three were men and three were women. None of these patients had clinical evidence of infection at the time of the study and none had diabetes mellitus.

Leukocyte Cell Preparations Approximately 10 mL of heparinized whole blood was obtained at each time point from the efferent line. At 15-minute sampling time, the volume of blood was increased to 20 mL. The blood was immediately cooled to 4°C, and centrifuged at 2,000 x g (1,000 rpm) for 10 minutes at 4°C in a refrigerated centrifuge. The platelet-rich plasma was discarded. One normal (IN) Tris buffer (0.025 mollL Tris, 0.12 mol/L NaCl, 0.005 mollL KCl, pH 7) was then added to make up the original volume. This was then layered over 1% dextran and allowed to sediment for 60 minutes at 4°C. The nonagglutinated aspirate was collected and then washed twice with 'llis buffer. Hypotonic lysis of residual red blood cells was performed using 0.2N Tris at 1°C for 45 seconds. The cells were then washed in Tris buffer and a second hypotonic lysis was performed. The cells were then washed three times with the last two washes performed in IN Tris plus 1.2 mmollL CaCl 2 and 1.0 mmollL MgCI 2 • They were subsequently counted and adjusted to a final concentration of 2.5 x 10' cells/mL.

Intracellular Dye-Loading With DCFH-DA A stock solution of DCFH-DA at a concentration of 5 mmoll L in ethanol was prepared and 1 I'L was added to each 1 mL of buffy coat cell preparation containing 2.5 x 10' cells; the cells were then incubated for 15 minutes at 37°C in a shaking water bath to allow diffusion inside the cells and were then immediately placed on ice. These were then assayed by flow cytometry as described below. In experiments where agonists' response was measured, the cell preparations were rewarmed to 3]oC in a shaking water bath for 5 minutes, the agonist (eg, C5a or F-Met-Leu-Phe [FMLP]) added, and the cells kept at 3]oC for another 25 minutes to allow for cell activation. They were then placed at O°C and assayed using flow cytometry within less than 4 hours. Preliminary studies showed that fluorescence signal intensity was stable under these conditions for a minimum of 4 hours from the time the reaction is quenched at O°C (data now shown).

Flow Cytometric Analysis of DCF Fluorescence All analyses were performed using an Ortho 2150 dual laser flow cytometer (Ortho Flow Systems, Westwood, MA). Using a combination of forward-angle light scatter and 90 a light scatter, gating of the flow cytometer was achieved for both the granulocyte and monocyte cell populations separately. Standardization of the instrument for particle size and fluorescence was achieved using calibrated beads (Flow Cytometry Standards, Research Triangle Park). In all cases, samples were run as duplicates and at least 2,000 cells within the combined granulocyte and monocyte gates were counted.

Hemodialysis Studies All patients were studied using first-use cuprophane hemodialysis membranes. A bicarbonate dialysate (Na, 140 mEq/L; K, 2 MEq/L; Ca, 3.5 mEq/L; HC0 3 , 30 mEq/L; acetate, 3 mEq/L) was used in all cases. Blood samples were drawn predialysis and subsequently from the efferent line to the dialyzer

HIMMELFARB, LAZARUS, AND HAKIM

at 15, 30, 60, and 240 minutes after initiation of dialysis. Anticoagulation was maintained using heparin at an initial rate of 50 U/kg, with additional doses given if activated clotting time was less than 1.5 times predialysis values.

Statistics All statistical analyses were performed using one-tailed paired Student's t tests analysis, and statistical significance was assumed for P :s 0.05.

RESULTS

To define optimal conditions for the use of DCFH-DA dye, we first studied the concentration and time dependence of activation of normal human granulocytes in response to standard neutrophil agonist C5a (a generous gift of Drs Dennis Chenoweth and Norma Gerard) and FMLP. Studies with C5a demonstrated that concentrations of C5a at or above 2 x 10- 8 mollL (up to 1 x 1O- 7mollL), the production ofROS was near maximal. Thus, at 2 x 10- 8 mollL, mean channel fluorescence was 155 ± 35 channels, while at 1 x 10 -7 mollL C5a, mean channel fluorescence was 172 ± 7.5 (Fig 1). Similar studies with FMLP demonstrated that peak H 20 2 production occurred at 10- 7 mollL where mean channel fluorescence was 278 ± 2.6. Generation of H 20 2 at either 10-6 or 10- 8 mollL FMLP was slightly less at 209 ± 28 and 153 ± 108 channels, respectively. In subsequent studies, the concentration of agonist used was 2 x 10- 8 mollL for C5a and 1 x 10-7 mollL for FMLP. Additional studies were performed using C5a to establish appropriate assay times for determining

20

-::;;

w

'"+1150 ...J

W

Z Z

« I

u 100

w u

z w

u

'"a::w g...J

50

LL

Z

«

w

::;;

o

III

10. 10

I

1

I

I

'/< I

III

10-8

[

I

t

I

2x10- 8

I

I

5x10·8

I

10- 7

C5A DOSE (nM]

Fig 1. Dose-response curve of DCF fluorescence by neutrophils to C5a (in vitro).

273

ROS DURING DIALYSIS

:::!:

UJ

300

(f)

+1 CI>

0:::: 0::::

250

X! !

C

.c U

CI>

u

200

0::::

CI> on CI> 0

u

....

150

::l

u::0:::: C

Q.I

:::!:

100

0

5

/r!---1

10

15

20

Time (min)

Fig 2. In vitro time course of DCF fluorescence of neutrophils: ., in response to 2 x 10- 8 C5a; 0, without addition of C5a.

ROS production. Studies with 2 x 10- 8 mollL C5a demonstrated a plateau effect that was evident by 20 to 30 minutes after addition of agonist (Fig 2). Fifteen minutes after stimulation with C5a, the fluorescence response was 78.3 % of the maximal response obtained at 30 minutes. Under the same conditions, ROS increased minimally when no C5a was added (Fig 2). The dialysis-induced generation of granulocyte ROS was studied in blood samples harvested during dialysis. Fifteen minutes after the initiation of dialysis, there was an increase in mean channel fluorescence from 191 ± 30 to 263 ± 33 fluorescence channels (P < 0.01) (Fig 3). Thirty minutes after the initiation of hemodialysis, mean channel fluorescence of granulocytes had declined to 218 ± 28 fluorescence channels. These levels re~

mained nearly constant at 60 minutes (217 ± 32 fluorescence channels), but were not statistically significantly higher than predialysis values (P = NS). At the termination of dialysis, mean fluorescence channels had decreased to 180 ± 26 fluorescence channels (P = NS). Figure 4 represents data obtained during the same intradialytic studies by gating on the monocyte population. At 15 minutes after the initiation of hemodialysis, ROS-induced fluorescence of DCF in monocytes had increased from 157 ± 37 predialysis to 182 ± 36 (P < 0.05). Thirty minutes after initiation of hemodialysis, ROS-induced fluorescence in these monocytes had decreased to 165 ± 28 channels (P = NS from baseline). By the end of the hemodialysis session, mean channel fluorescence had returned to baseline values (158 ± 27 fluorescence channels). Thus, in a fashion similar to neutrophils, monocytes isolated during cuprophane hemodialysis show a statistically significant increase in intracellular hydrogen peroxide 15 minutes after the initiation of hemodialysis. The use of flow cytometry for measurement of the production of ROS allows the assessment of ROS production on a cell-by-cell basis, as well as their distribution in a population of cells. In four of the six patients at all time points, as well as in normal controls and nondialyzed uremic patients, the distribution of ROS production by granulocytes was unimodal and gaussian in distribution. In two of the six patients studied, a bimodal peak of fluorescence was observed in the IS-minute PMNL sample as compared with predialysis samples. This bimodal peak persisted at 30 minutes,

200Mi

UJ

~ 250 [

+1

(f)

Q.I

Qi

.c

.c

w +,

(f)

0:::: 0:::: C

0:::: 0:::: C

U

U

Q.I

Q.I

u

u

Q.I

Q.I

0::::

u on

~

0 .;!

u.

0:::: C

Q.I

~

0::::

u on

l50t 100

0

~ 0

ltltp
30

I

60

I

90

II

.2

U.

I

240

DIALYSIS TIME (min)

Fig 3. Change in ROS-specific fluorescence of neutrophils during dialysis with cuprophane membrane (in vivo).

0::::

c

Q.I

~

150

':t 00

it

f * p
I

30

I

60

I

90

I~

240

Dialysis Time (min)

Fig 4. Change in ROS-specific fluorescence of monocytes during dialysis with cuprophane membrane (in vivo).

274

HIMMELFARB, LAZARUS, AND HAKIM IBB

100 ~ 0

z

-l -l

9B

PRE

19

z

79

-l -l

79

w 68 u S8 w > 49

w 69 u 58 w > 49

t-

t-

a: J8 -l

w 29

0<:

19

-l

w 29 ~

\ 2B

49

IB9 129 148 69 69 FLUORESCENCE

168

168

-l -l

0

78

-l

0<:

18 BB

29

B

F-J \ 49

69 69 ISS 129 148 FLUORESCENCE

168

169

298

IBB

299

613 MINS

S9

z -l -l

79

w 60 u 58 w > 4B

w 69 u 50 w > 40 w 2B

...

99

30 MINS

8B

tor JB

\

1ge

98

z

19 9

288

IBe 0

(\

a: J9

\

99

15 MINS

S9

0

!",

;I

' 28

.

4B

•

\,

tor J8

-l

I

I

IS9 129 149 66 89 FLUORESCENCE

I

169

.169

w 2B 0<: IB

I

288

J/'

.

B B

2B

.

4B

\\ .\

"'"

...

66 BB IBB 129 14B FLUORESCENCE

16B

Fig 5. Distribution of ROS-Induced fluorescence In PMNL during cuprophane hemodialysis In an individual patient. A bimodal peak Is clearly seen at 15 minutes, but is less evident by 60 minutes after initiation of dialysis.

but was less evident 60 minutes after initiation of dialysis (Fig 5). This suggests that overall mean DCF fluorescence data shown in Figs 3 and 4 may have underestimated the extent of the production of ROS by a subset of more activated neutrophils. In the next series of experiments, granulocytes harvested throughout the hemodialysis process were isolated and incubated with 2 x 10- 8 C5a as described above to assess their ability to respond to further stimuli. The intensity of DCF fluorescence in response to the standard (in vitro) dose of C5a was compared at each time point to the amount of "baseline" intradialytic fluorescence obtained at the same time point. Prior to the initiation of dialysis, the addition of 2 x 10- 8 mollL C5a resulted in a 22.8% ± 5.1 % increase in the fluorescence of the DCFH-DA-Ioaded cells. Fifteen minutes after the initiation of dialysis, the addition of exogenous C5a resulted in only a 7.2% ± 2.2% increase in DCF fluorescence. This decreased responsiveness to exogenous C5a persisted at 30 and 60 minutes, respectively (7.7% ± 2.3% and 9.8% ± 2.5%), and was statistically

significant at the three time points (P < 0.05) (Fig 6). However, granulocytes harvested at the end of dialysis were as responsive to C5a as granulocytes harvested predialysis (22.2% ± 4.3% increase in fluorescence at the end of dialysis v 22.8% ±

5

o

~--~L---~----~--1r-l-

o

30

60

90

240

DiolysisTime (min) Fig 6. Percent increase In ROS-speclflc fluorescence of neutrophils harvested Intradlalytically to further stimulation by C5a (2 x 10- 8 moIlL).

275

ROS DURING DIALYSIS OJ

c:

40

OJ

Ul

C

co

....

~

30

o

OJ

u~

~~ 20 ~(J)

g-+. OJ

Li:

10

'5 OJ

Ul

C

~ c: 'cf'.

OL--~--L------'----L--I~

o

30

60

90

240

Dialysis Time (min)

Fig 7. Percent increase in ROS-specific fluorescence of neutrophils harvested intradialytically to further stimulation by FMLP (1 x 10- 7 molll).

5.1 % stimulation at the beginning of dialysis) (P = NS). Thus, the responsiveness of granulocytes harvested during dialysis to exogenous C5a was diminished through at least the first 60 minutes of hemodialysis with cuprophane membranes. The ability of granulocytes harvested during dialysis to respond to exogenous FMLP was similarly reduced (Fig 7). Before the initiation of dialysis, the addition of 1 x 10- 7 mollL FMLP resulted in a 27 % increase in granulocyte hydrogen peroxide production. Fifteen minutes after the initiation of hemodialysis, granulocyte responsiveness to FMLP was significantly reduced (P < 0.05) to an increase of 5 % ± 2 % over baseline. At 30 and 60 minutes after the initiation of dialysis, granulocyte responsiveness to exogenous FMLP remained lower than predialysis, at 13 % ± 6% and 14% ± 8%, respectively, but these were not statistically significant. By the end of the hemodialysis sessions, mean response to FMLP by PMNL had returned to baseline values. DISCUSSION

In this study, we demonstrated that both granulocytes and monocytes harvested after the initiation of dialysis with cuprophane membrane have a transient but significant increase in intracellular ROS production as compared with cells harvested predialysis. The peak production of ROS during dialysis corresponds to the time of maximum complement activation during dialysis with this membrane. l1 However, it is likely that the measured fluorescence of ROS production reported in this study is an underestimate of the true extent of ROS

production because of the lapse of time between the harvesting of PMNL and their incubation with DCFH-DA. Intracellular enzymes such as catalase may act to rapidly degrade these powerful oxidants. In this study we were also able to demonstrate increased intracellular ROS (as an index of activation) in the monocyte population. While there have been relatively few studies of this important cell population during dialysis, there is increasing interest in defining whether or not monocyte-derived IL-l plays a role in dialysis-related signs and symptoms.1.2 The demonstration in this study of monocyte activation lends indirect support to this hypothesis; however, the study did not demonstrate that monocyte activation contributes to dialysis-related signs and symptoms such as fever or hypotension. We also examined the effect of hemodialysis with cuprophane membranes on the responsiveness of leukocytes to further stimulation with ex0genous C5a and FMLP. Our data demonstrate that PMNL harvested during dialysis have a depressed responsiveness to both C5a and FMLP intradialytically. This may have bearing on the rate of infection, which is a leading cause of morbidity and mortality in long-term dialysis patients. 12-1S Many of these infections are directly attributable to transient bacteremias occurring early during the dialysis procedure, with Staphylococcus aureus as the most commonly involved pathogen. 12- 16 The decreased ability of neutrophils harvested during dialysis to respond to FMLP (an analog of a bacterial cell wall) or to C5a, a product of the inflammatory response to most of these pathogens, may be a factor in the increased susceptibility to infection in these patients. Although in this study we used the intracellular production of ROS as an index of cell activation, 16 it is important to keep in mind that numerous studies have pointed out the deleterious effects of extracellular ROS production, which range from endothelial damage and acceleration of atherosclerosis to induction of DNA strand breaks and carcinogenesis. 17-28 Whether repetitive ROS production during dialysis with complement-activating membranes causes any of these effects in a dialysis patient population remains speculative. While this study attempts to link complement activation to intracellular PMNL and monocyte ROS production, it must be noted that only cuprophane dialysis membranes were studied. Crossover stud-

276

HIMMELFARB, LAZARUS, AND HAKIM

ies using this methodology with less complementactivating membranes will further define this relationship; similarly, the potential role of endotoxin transfer from the dialysate in membranes with more open structures, such as high flux membranes, may be studied using this methodology. In summary, we have demonstrated that hemodialysis with cuprophane membranes causes increased intracellular production of ROS in both leukocytes and monocytes, and confirms their activation during dialysis with this membrane. This repetitive production of ROS may contribute to conditions as diverse as malignancy, and athero-

sclerosis in hemodialysis patients. Furthermore, additional responsiveness of leukocytes during dialysis to either C5a or FMLP is downregulated through at least the first hour of dialysis. This observation may be important in the pathogenesis of bacteremic infections in hemodialysis patients. ACKNOWLEDGMENT We thank Dr Douglas T. Fearon for advice, and Drs Dennis E. Chenoweth and Norma Gerard for their generous gift of C5a. We thank Cynthia Perzanowsky and Robert Hoffman for technical assistance. Our thanks also to Karen A. Kinne and Annie Bernard for help with preparation of the manuscript.

REFERENCES 1. Henderson LW, Koch KM, Dinarello CA, et al: Hemodialysis hypotension: The interleukin hypothesis. Blood Purif 1:3-8, 1983 2. Shaldon S: The interleukin hypothesis: A reappraisal after 6 years. Semin Dial 2:163-172, 1989 3. Blumenstein M, Ziegler-Heitbrock HWL, Schiller B, et al: Differential activation of monocytes in haemodialysis patients exposed to different types of membranes. Scand J Immunol 31:183-190, 1990 4. Haeffner-Cavaillon N, Cavaillon JM, Ciancioni C, et al: In-vivo induction of interleukin-I during hemodialysis. Kidney Int 35: 1212-1218, 1989 5. Herberlin A, Nguyen AT, Zingraff J, et al: Influence of uremia and hemodialysis on circulating interleukin-l and tumor necrosis factor ex. Kidney Int 37: 116-125, 1990 6. Schindler R, Lonnemann G, Miller LC, et al: Transcriptional activation but not synthesis of interleukin-l and tumor necrosis factor by complement activation in the absence of endotoxin. Kidney Int 37:319, 1990 7. Lonnemann G, Bingel M, Floege J, et al: Detection of endotoxin-like interleukin-l-inducing activity during in vitro dialysis. Kidney Int 33:29-35, 1988 8. Luger A, Kovarik J, Stummvoll HK, et al: Blood-membrane interaction in hemodialysis leads to increased cytokine production. Kidney Int 32:84-88, 1987 9. Betz M, Haensch GM, Rauterberg EW, et al: Cuproammonium membranes stimulate interleukin-l release and arachidonic acid metabolism in monocytes in the absence of complement. Kidney Int 34:67-73, 1988 10. Bass DA, Parce Jw, Dechatelet LR, et al: Flow cytometric studies of oxidative product formation by neutrophils: A graded response to membrane stimulation. J Immunol 130: 1910-1917, 1983 II. Hakim RM, Breillatt J, Lazarus JM, et al: Complement activation and hypersensitivity reactions to dialysis membranes. N Engl J Med 311:878-882, 1984 12. Yu VS, Geotz A, Wagener M, et al: Staphylococcus aureus nasal carriage and infection in patients on hemodialysis: Efficacy of antibiotic prophylaxis. N Engl J Med 315:91-96, 1986 13. Nsouli KA, Lazarus JM, Schoenbaum SC, et al: Bacteremic infection in hemodialysis. Arch Intern Med 139:12551258, 1980 14. Siddiqui JY, Fitz AE, Lawton RL, et al: Causes of death

in patients receiving long-term hemodialysis. JAMA 212: 13501354, 1970 15. Quarles LD, Rutski EA, Rostand SG: Staphylococcus aureus bacteremia in patients on chronic hemodialysis. Am J Kidney Dis 6:412-419, 1985 16. Nguyen AT, Lethia C, Zingraff J, et al: Hemodialysis membrane-induced activation of phagocyte oxidative metabolism detected in vivo and in vitro with microamounts of whole blood. Kidney Int 28:158-167, 1985 17. Sacks T, Moldow CF, Craddock PR, et al: Oxygen radicals mediate endothelial cell damage by complement stimulated granulocytes: An in-vitro model of immune vascular damage. J Clin Invest 61:1161-1167, 1978 18. Weitberg AB, Weitzman SA, Clark EP, et al: Effects of antioxidants on oxidant-induced sister chromatid exchange formation. J Clin Invest 75:1835-1841, 1985 19. Schraufstiitter I, Hyslop PA, Jackson JH, et al: Oxidantinduced DNA damage of target cells. J Clin Invest 82:10401050, 1988 20. Weitzman SA, Weitberg AB, Clark EP, et al: Phagocytes as carcinogens: Malignant transformation produced by human neutrophils. Science 227:1231-1233, 1985 21. Weiss SJ, Young J, LoBuglio AF, et al: Role of hydrogen peroxide in neutrophil-mediated destruction of cultured endothelial cells. J Clin Invest 68:714-721, 1981 22. Harlan JM, Killen PD, Harker LA, et al: Neutrophilmediated endothelial injury in vitro: Mechanisms of cell detachment. J Clin Invest 68:1394-1403, 1981 23. Harlan JM: Leukocyte-endothelial interactions. Blood 65:513-525, 1985 24. Britigan BE, Cohen MS, Rosen GM: Hydroxyl radical formation in neutrophils. N Engl J Med 318:858-859, 1988 25. Birnboim HC: DNA strand breakage in human leukocytes exposed to a tumor promoter, phorbol myristate acetate. Scinece 215:1247-1249, 1982 26. Lindner A, Farewell VT, Sherrard DJ: High incidence of neoplasia in uremic patients receiving long-term dialysis. Nephron 27:292-296, 1981 27. Lewis MS, Whatley RE, Cain P, et al: Hydrogen peroxide stimulates the synthesis of platelet-activating factor by endothelium and induces endothelial cell-dependent neutrophil adhesion. J Clin Invest 82:2045-2055, 1988 28. Nathan CF: Neutrophil activation on biological surfaces. J Clin Invest 80: 1550-1560, 1987