Comparison of Eosinophilia in Patients Undergoing Peritoneal Dialysis and Hemodialysis

Comparison of Eosinophilia in Patients Undergoing Peritoneal Dialysis and Hemodialysis Rebecca Backenroth, MD, MPH, Bruce S. Spinowitz, MD, Marilyn Galler, MD, Ronald A. Golden, MD, Joel H. Rascoff, MD, FACp, and Chaim Charytan, MD, FACP • Eosinophilia (E) has been noted in hemodialysis (HD) patients, but its etiology is not clear. In an effort to clarify this phenomenon, we prospectively studied patients initiating dialysis in our outpatient HD and peritoneal dialysis programs. Rate of E was greatest for a small group of four continuous cycling peritoneal dialysis patients (75%), less for 63 HD patients (41 %), and least for 66 continuous ambulatory peritoneal dialysis (CAPO) patients (21 %, P < .05, HD v CAPO). Increasing E rates among the groups paralleled increased frequency of tubing changes. There were no differences in etiology of renal disease, medications, types of dialyzers, types of access, or transfusion frequency that could account for the E. IgE levels did not correlate with E. The data suggest that the dialysis procedure or the tubing changes may be causing the E, but the possibility that uremia, itself, is important in the pathogenesis of dialysis E is also discussed. © 1986 by the National Kidney Foundation, Inc. INDEX WORDS: Eosinophilia; hemodialysis; CAPO; allergy; uremia.

E

OSINOPHILIA (E) has been noted in hemodialysis (HD) patients by several investigators. 1-9 It has been associated with an increase in allergic reactions,2 and at times, with severe lifethreatening hypersensitivity reactions. 5 In some patients, eosinophil counts have been as high as in the hypereosinophilic syndrome. 1O The cause of the E is controversial and may be multifactorial. The E has been attributed to dialyzer reuse,1 but it is clearly present in patients not exposed to such practices.V2- 3 In another study, E and hypersensitivity reactions seemed related to Cuprophan dialyzers (Travenol Laboratories Inc, Deerfield, Ill) and cleared with use of nonCuprophan dialyzers 4 ; however, we have noted a similar incidence of E in patients using Cuprophan and nonCuprophan dialyzers. 2 Still others have found a relation between ethylene oxide (ETO) gas sterilization of equipment and dialysis E.8 In addition, we and others have also noted E in peritoneal dialysis (PD) patients. 11-13 In an effort to delineate the possible etiology of E in dialysis patients, we undertook a prospective study of all persons entering our outpatient HD and PD programs. From the Renal Division, Department of Medicine, Booth Memorial Medical Center, Flushing, New York. Part of the data herein was presented in abstract form at the Hypersensitivity in Dialysis Conference, Louisville, Kentucky, July 21-22, 1983. Address reprint requests to Bruce S. Spinowitz, MD, Booth Memorial Medical Center, Nephrology Department, 56-45 Main St, Flushing, NY 11355. © 1986 by the National Kidney Foundation, Inc. 0272-6386/86/0803-0008$03.00/0

186

MATERIALS AND METHODS

Patient Selection All patients who began outpatient treatment in our Satellite Facility (Flushing, NY) between May 1981 and Dec 1982 were included in the study, unless they stayed in the program for less than 2 months, had a recognized explanation for E, or were receiving steroid therapy. All patients had inpatient dialysis before admission to the outpatient unit.

Laboratory Tests All patients had a complete blood count with differential on admission to the dialysis unit, and three times per year thereafter, ie, in January, May, and September. They were done on a Coulter F (Coulter Electronics Inc, Hialeah, Fla) and Technicon Hemolog D (Technicon Instruments Corp, Tarrytown, NY). Differential was reported on 10,000 WBCs. Absolute eosinophil counts were performed manually in 3 different months of the year, ie, in February, June, and October. Occasional eosinophil counts were verified in two different laboratories. There was also good agreement between 66 eosinophil counts determined by differential, as well as by absolute count (r = .81). IgE levels were performed by SmithKline Clinical Laboratories, Inc, using the Kallestad radioimmunoassay technique (Kallestad Laboratories, Inc, Austin, Texas). Stool was checked for ova and parasites when E occurred on more than one determination.

Dialysis Procedure Continuous ambulatory peritoneal dialysis (CAPD) patients had four daily exchanges with autoclaved, lactate-based Dianeal (Travenol Laboratories) bags and had one tubing change per month. Continuous cycling peritoneal dialysis (CCPD) patients had a similar number of exchanges with similar bags, but had daily tubing changes. HD patients were dialyzed with acetate-based dialysate for four hours three times a week. A variety of non-reused dialyzers, including Cuprophan, regenerated cellulose, and saponified cellulose ester, were used. Dialyzers and all tubing sets were ETO-sterilized.

American Journal of Kidney Diseases, Vol VIII, No 3 (September), 1986: pp 186-191

EOSINOPHILIA IN PO AND HD

187

Definitions

•.•••• f'Cr1.tt'nl6J dllllyab pat.lent "'hl1 el("Vdt,OO ooairqlhll CXUI 'a) toof1O Me! lYB1& within a l"'Dnth forn t.hc! C!WltUllibld oeutt.la).

E was defined as one or more episodes of greater than 5 % of WBC, or 399 eosinophilsl/LL. Persistent E was defined as three consecutive elevated counts not reversed at a later date. Allergies were defined as a history of rash or anaphylactic reaction to medications or contrast agents. All charts were reviewed by a physician (RB) for possible etiologies of E. Statistical methods included nonpaired t test and chi-square analysis. Results were expressed as means ± SEM .

1500

1000

RESULTS

CAPD

Sixty-six patients were eligible for the study; 39 were male and 27, female. Forty-one were white; 19, black; 4, Hispanic; and 2 Oriental. The average age was 65 years and duration on dialysis averaged 9 months (Table 1). Nineteen CAPO patients had E (Table 1, Figs 1 and 2). However, five of these patients always had some HO treatments within 1 month before each elevated eosinophil count. The exclusion of these five patients from analysis increases the difference in incidence of E between CAPO (21 %) and HO (41.3 %) to a statistically significant level (P < .05). Only two CAPO patients had persistent E. In most patients, E developed within the first 6 months of CAPO, with only two patients developing E later (Table 2). Allergies were no more frequent in E patients than non-E patients (32 % v 30%). Transfusions were given to 16.7% ofE and non-E patients, and these were not temporally related to the episodes of E. IgE levels were higher in those with E and higher in CAPO than HO, but not to statistically significant level (Tables 1 and 3). Since a single-type catheter (Toronto-Western, Toronto, Canada) was used in 61 of the 66 patients, its role in E could not be assessed. Table 1.

(!)

500

*Means

±

SEM.

tP < .05, CAPO v HD.

:j:Normal < 122 U.

i....

1 ~

A

'f

......

Nonml

.~~

1

-,:-

-:

~

100

tmmIALYSIS

CAPO


Fig 1. Mean eosinophil counts of patients on HO, CAPO, and CCPO. Circled dots represent patients with HO in month prior to elevated E count.

(J)

I-

Z

W

I
a.

u.

o

at

50r------------------------------, 40 30

20

w

ClI

5z

10

e

2 3 4 5 6 7 8 0 NUt'lBER OF ELEUATED COUNTS . . CAPD N=66 ~ HD N=63

'3

Fig 2. Persistence of eosinophilia in 129 CAPO and HD patients on 690 determinations.

Patients on CAPO, HO, and CCPO

HO

CAPO (n = 66)

Age in years (range) Months on dialysis (range) No. with eosinophilia (%) No. with eosinophilia and without HD in previous month (%) Eosinophil counts (range) IgE:j: (no. tested) Tubing changes per month Dialysate base

:- - -

-l

55 ± 2* (20-79) 9 ± 0.7* (2-21) 19 (28) 14 (21)t 216 153

± ±

19* (0-5317) 41* (36) 1 Lactate

(n

=

CCPO (n = 4)

63)

60 10

± ±

2 (28-89) 0.6 (2-24) 26 (41)

253 109

± ±

30 (0-1566) 18(54) 13 Acetate

52 ± 4 (41-57) 7 ± 0.5 (5-7) 3 (75)

303 17

± ±

62 (0-816) 6 (2) 30 Lactate

188

BACKENROTH ET AL Table 2. Duration in Dialysis Program (mo)

0-3 4-6 7-9 10-12 13-15 16-18 19-21 22-24

Timing of Onset of Eosinophilia CAPD

HD

CCPD

ElTotal at Risk (%)

ElTotal at Risk (%)

ElTotal at Risk (%)

13*/66 (20) 4/38 (11) 1/25 (4) 0/16 (0) 0/10 (0) 017 (0)

116 (17) (0)

12/63 7/50 3/34 3/20 1/11 0/8 0/2

(19) (14) (9)

2/4 (50) 1/2 (50)

(15) (9) (0) (0)

011

* Five of the 13 had HD within 1 month before elevated eosinophil count.

Presumed etiologies of renal failure are presented in Fig 3. Fewer diabetics were in the E group, but this is not statistically significant. Medication use (Fig 4) was similar in the E and non-E groups , but was different in CAPD and HD groups, in that significantly more HD patients were on anabolic steroids (intramuscular nandrolene decanoate, Deca-Durabolin; Organon Pharmaceuticals, West Orange, NJ) and parenteral iron dextran (lmferon, Merrill Dow Pharmaceuticals Inc, Cincinnati, Ohio). It is of interest that four of the 19 patients with E (21 %) also had one episode each of culture-negative eosinophilic peritonitis 13 while only 14 of the 164 patients in the entire CAPD program had eosinophilic peritonitis (8.5 %, P value NS). The occurrence of peripheral E and peritoneal E were not, however, temporally related in the former four patients. HD Patients Sixty-three eligible patients were on dialysis an average of 10 months; 29 were female and 34, male. Average age was 60 years (Table 1). There were 36 white , 23 black, and 4 Hispanic patients (Table 1). Twenty-one patients had E (Figs 1 and 2), but only four demonstrated persistent E. The E developed in the first 6 months in most patients (Table 2). More E patients (38 %) demonstrated allergies than non-E patients (24 %), but Table 3.

•

*Means ± SEM . tlgE > 122 U.

Jo£J1ODIf't.YSJS

-. ~

•

E ..-:it.

0

IUI-£

CCH

,,-n

~!-L~--,.cliTf1 0 •

[Ii}

D'I(J,["

il; I

Fig 3. Etiology of end-stage renal disease in the three groups of dialysis patients. Abbreviations: SCLER, nephrosclerosis; KW, diabetic nephrosclerosis; CGN, glomerulonephritis; CIN, chronic interstitial nephritis; PKD, polycystic kidney disease.

Mean Ig Levels in Dialysis Patients With and Without Eosinophilia CAPD

IgE Patients tested With elevated IgEt (%)

r:J 'U~-E 1r--'1

E H.I~

HD

CCPD

E

Non-E

E

Non·E

E

Non·E

196 ± 83* 12 42

131 ± 45 24 21

143 ± 32 22 41

86 ± 21 32 22

23 1 0

11 1

o

189

EOSINOPHILIA IN PD AND HD CI1f'D ':

;"

> ~

":!:'" "i ~

<: J w ~

7~

6u

~W

~.(1

::0

a

w

C>:

u.. w ::>

:lli

::Cl

t-

1&


-' W

U

•

""

E 1" I ~

. . E N=26

!..J

u 5{,

§

'"

to.

~ :;: J

"'""

b 11 t1£.LdC"11(lt1,,, t:l ""WE H- J1

~

•

I22a

NON-E 11=37

HEI10D I nL YS I S

':

;..

TYPE OF DI ALYZER

E I,",,~

Fig 5. Dialyzer types used in HD patients during or within 1 months before eosinophil determinations: CF12-CF12.11 and CF15-CF15.11, Cuprophan (Travenol); CD13-C.D. 1.3 and CD18-C.D. 1.8, regenerated cellulose (Cordis Dow); CD90-C.D. 90 and CD35-C.D. 135, saponified cellulose ester (Co rid Dow), and OTHR (others). Totals exceeding 100% per group reflect multiple dialyzers used In individual patients.

CCPD Fig 4. Representative medications in CAPD and HD patients. Abbreviations: A, Basaljel (Wyeth Laboratories, Philadelphia, Pa); B, other binders; C, calcitrlol; D, dihydrotrachysterol; E, antipruritics; F, digoxin; G, nitrates, H, /3-blockers; I, hydralazine, J, other antihypertensives; K, cimetidlne; L, Insulin; M, anabolic steroids; N, parenteral Iron dextran.

this was not statistically significant. Transfusions were given to 46 % of E patients and 49 % of non-E patients, and were not temporally related to the E. IgE levels were higher in the Ethan non-E patients (Table 3), but HD patients had lower overall IgE levels than CAPD patients (Table 1). '!Ypes of vascular access were similar in the E and non-E groups, including Gore-Tex (Elkton, Md) bovine and Hemasite grafts (Renal Systems, Inc, Minneapolis, Minn) . Dialyzer types used at the time of or within 1 month prior to the eosinophil counts were not markedly different in the two groups (Fig 5). Changes in dialyzers were temporally associated with onset of E in seven patients; it was associated with a change to Cuprophan (Travenol CF 1211) dialyzers in four patients, to regenerated cellulose (Cordis Dow 1.3; Cordis Dow, Miami Lakes, Fla) in two patients, and to saponified cellulose ester (Cordis Dow 125) in one patient. Presumed etiologies of renal disease are presented in Fig 3. As in the CAPD group, the excess of diabetics in the non-E group did not reach statistical significance. Medication use was similar in the E and non-E groups, as seen in Fig 4.

There were only four eligible CCPD patients, who were on dialysis for an average of 7 months. Average age was 62 years (Table 1). Three were male and 1, female; 3 were white and 1, black. Etiology of renal failure is presented in Fig 2. Three of the four patients had E and in one, it was persistent.

All Groups E was not seasonal in any subgroup and seemed unrelated to age, sex, and race. In analyzing the incidence of E in the three dialysis modalities, it appears that the incidence of E increased relative to the frequency of tubing changes (Table 1). Thus, in CAPD with one monthly tubing change, the rate of E was lowest at 21 % to 28 %; in HD with tubing changes three times a week, the rate was 41 %; whereas in CCPD with daily tubing changes the rate was highest at 75 % (r = 1, P < .001). DISCUSSION

This prospective study shows that HD patients have a higher mean eosinophil count and a higher frequency of Ethan CAPD patients. This is similar to the findings of Dratwa et al 4 who found E in 45% of 62 HD patients, but only 8% E in 12 CAPD patients and 10% E in normal controls. The high incidence of E in our CCPD patients is intriguing; however, it must be interpreted with caution since the CCPD group was very small.

190

The E in our study seems unrelated to transfusions, specific dialyzers, or seasons. It does not appear to be related to medications, despite the higher use in the HD group of intramuscular anabolic steroids and iron dextran, since, in fact, within the HD group, fewer of these agents were used in the E group. The etiology of dialysis E may be related to the dialysis procedure or the uremic state per se. There are several lines of evidence incriminating the dialysis procedure as the cause of E: (1) the lack of E in uremics not yet on dialysis 4.6; (2) the acute rise of eosinophil counts during individual HD treatments, even in patients who have no predialysis E9.14; and (3) the apparent association of E with formaldehyde sterilization of reused dialyzers, I Cuprophan dialyzers,5 and ETO sterilization of dialysis equipment. S Our finding of different E rates in different dialysis modalities is consistent with a procedurerelated etiology of E. The increasing incidence of E in groups with increasing frequency of tubing changes, ie, CCPD > HD > CAPD, suggests that the tubing changes may be related to the E. The sensitizing agent could be the ETO used to sterilize tubing and dialyzers, as it has been demonstrated to cause allergic reactions, positive skin tests, and increased E, as well as a positive radioallergosorbent test (RAST) for ETO in HD patients. S.15.16 Differential exposure to other agents such as heparin or plasticizers could also explain the varying degrees of E among the different dialysis modalities; however, we were unable to quantitate such exposure in our patients. There is also data in the literature suggesting that abnormal eosinophil homeostasis is intrinsic to the uremic state and may be independent of dialysis. Gabizon et al 6 found bone marrow E in nondialyzed uremics in whom the degree of marrow E correlated with serum creatinine levels. These patients did not, however, have peripheral E. Similarly, 35 nondialyzed uremics studied by Dratwa et al 4 also did not have peripheral E, but had high IgE levels. Eosinophil cationic protein, an indicator of eosinophil turnover, has been found to be elevated in CAPD, HD, and nondialyzed uremics, all of whom did not have peripheral E.17 This again confirms abnormal eosinophil kinetics and suggests increased eosinophil turnover in uremia that is independent of dialysis. The presence

BACKENROTH ET AL

of peripheral E in dialyzed patients may, therefore, be the manifestation of a more severe degree of uremic disturbance in eosinophil homeostasis that is already present before dialysis and is only aggravated by HD, to a greater extent than with CAPD. The E, for instance, may be related to the immune dysfunction of uremic patients, which may, in turn, be affected differently by the different dialysis regimens. This possibility is supported by several lines of evidence: it is known that E is seen in some immune deficiency diseases and that E is stimulated by T lymphocytes. IS T lymphocytes have been found, by Tielmans et al, 19 to differ in HD patients with and without E, in that E patients had deficient in vitro suppressor function and a higher percentage of T helper cells. Of interest, several investigators have suggested that CAPD is better than HD in controlling in vitro lymphocyte and T lymphocyte function,20-22 as well as in restoring skin test reactivity to PPD and DCNB.22 It is possible, therefore, that differential control of the uremic immune dysfunction accounts for the lesser E seen in nondialyzed uremic and CAPD patients, compared to HD patients. Another unanswered question in dialysis E is its relation to IgE levels. Dratwa et al 4 found IgE elevations in non-E nondialyzed uremics and in HD patients, but not in CAPD patients. Furthermore, IgE elevation has been shown to be associated with HD E by several authors.4.9.19.23 The low IgE levels in some of our E patients do not preclude such an association since IgE is a tissuebound globulin and its absence in serum may simply reflect tissue-binding during eosinophil activation. However, our data fail to show a correlation between IgE levels and E. The increased incidence of eosinoiphilic peritonitis in CAPD patients with E is intriguing. If it represents a true trend, it suggests that systemic tendency to E can become manifest peritoneally or that an allergen introduced peritoneally causes local, as well as systemic, E. In conclusion, we found E to be frequent in dialysis patients. The incidence of E was greatest in a very small group ofCCPD patients, somewhat less frequent in HD patients, and least common in CAPD patients. The E had no apparent clinical etiology and was substantial in some patients. Further studies are indicated to delineate the implications and possible etiologies of E.

191

EOSINOPHILIA IN PO AND HD

REFERENCES 1. Hoy WE, Cestero RVM: Eosinophilia in maintenance hemodialysis patients. J Dialysis 3:73-87, 1979 2. Spinowtiz BS, Simpson M, Manu P, et al: Dialysis eosinophilia. Trans Am Soc Artif Intern Organs XXVII:161-167, 1981 3. Montoliu J, Lopez-Pedret J, Andreau L, et al: Eosinophilia in patients undergoing dialysis. Br Med J 282:2098, 1981 4. Dratwa M, Tielmans C, Brenez D: IgE levels in patients with chronic renal failure. N Engl J Med 305:960-961, 1981 5. Michelson EA, Cohen L, Dankner RE, et al: Eosinophilia and pulmonary dysfunction during Cuprophan hemodialysis. Kidney Int 24:246-249, 1983 6. Gabizon D, Kaufman S, Shaked U, et al: Eosinophilia in uremia. Nephron 29:36-39, 1981 7. Scheuermann EH, Fassbinder W, Frei U, et al: Eosinophilia in hemodialysis. Contrib Nephrol 36:133-138, 1983 8. Takahashi S, Ohsima H, Watanabe H, et al: Eosinophilia observed in regular hemodialysis patients. Nephron 28: 154, 1981 (abstr) 9. Perez-Garcia R, Gurbindo C, Gurbindo MD, et al: Eosinophilia in hemodialysis: Implication of the IgE-basophil system. Nephrologie 35:271-272, 1983 10. Fauci AS, Harvey 18, Roberts WC, et al: The idiopathic hypereosinophilic syndrome. Ann Intern Med 97:78-92, 1982 11. Gokal R, Ramos JM, Ward MK, et al: Eosinophilic peritonitis in continuous ambulatory peritoneal dialysis. Clin Nephro1 15:328-330, 1981 12. Humayun HM, Ing TS, Daugirdas JT, et al: Peritoneal fluid eosinophilia in patients undergoing maintenance peritoneal dialysis. Arch Intern Med 141:1172-1173, 1981

13. Spinowitz BS, Golden RA, Rascoff JH, et a1: Eosinophilic peritontis, in Gahl M, Kessel M, Nolph KD (eds): Advances in Peritoneal Dialysis. Amsterdam, Excerpta Medica, 1981, pp 276-280 14. Bodner G, Peer G, Zakuth V, et al: Dialysis-induced eosinophilia. Nephron 32:63-66, 1982 15. Poothullil J, Shimizu A, Day RP, et al: Anaphylaxis from the product(s) of ethylene oxide gas. Ann Intern Med 82:58-60, 1975 16. Dolovich J, Bell B: Allergy to a product(s) of ethylene oxide gas. J Allergy Clin Immunol 62:30-32, 1978 17. Hallgren R, Grefberg N, Venge P: Elevated circulating levels of eosinophil cationic protein in uremia as signs of abnormal eosinophil homeostasis. Nephron 36:10-14, 1984 18. Weller PF, Goetzl EJ: The regulatory and effector roles of eosinophils. Adv Immunol 27:339-371, 1979 19. Tielmans C, Collart F, Schandene E, et al: Eosinophilia and supspressive T cell function in hemodialysis patients. Kidney lnt 23: 163, 1983 (abstr) 20. Langhoff E, Ladefoged J: Improved lymphocyte transformation in vitro of patients on CAPD dialysis. Kidney Int 24:411, 1983 (abstr) 21. Collart F, Tielmans C, Schandene L, et al: HD, CAPD and cellular immunity. Kidney Int 24:421-422, 1983 (abstr) 22. Giacchino F, Quarello F, Pellerey G, et al: Continuous ambulatory peritoneal dialysis improves immunodeficiency in uremic patients. Neprhon 35:209-210, 1983 23. Spinowtiz BS, Sankarapandian P, Chary tan C: Elevated serum IgE levels in hemodialysis patients with eosinophilia. Kidney Int 21:180, 1982 (abstr)