Nephrol Dial Transplant (1990) 5: 130-134 © 1990 European Dialysis and Transplant Association-European Renal Association
Nephrology Dialysis Transplantation
Original Article
J. R. Boelaert1, R. F. Daneels1, M. L. Schurgers1, E. G. Matthys1, B. Z. Gordts2 and H. W. Van Landuyt2 'Unit for Renal and Infectious Diseases and 2Department of Microbiology, Algemeen Ziekenhuis St Jan, Brugge, Belgium
Abstract. The incidence of bacteraemia in relation to the degree of transfusional iron overload was studied prospectively in patients from one haemodialysis unit over a 2-year period, with a total follow-up of 181.3 patient-years in 158 patients. Every 3 months, the patients were classified according to the serum ferritin in one of three groups: 1000 ug/1. Twenty-nine episodes of bacteraemia were recorded over 181.3 patient-years (yearly incidence of 0.160). The yearly incidence of bacteraemia was 0.1173 and 0.1101 for ferritin 1000 ug/1 group, the incidence was 0.3404 (P^. 0.005 versus the ferritin ^ 1000 ug/1 group). After stratification for patient's age (at inclusion in the study) and duration of haemodialysis therapy, the higher incidence of bacteraemia in the ferritin >1000 versus < 1000 ug/1 groups persisted (/>< 0.005). This prospective study confirms previous retrospective studies in showing that acquired transfusional iron overload in haemodialysis is associated with a greater risk of bacteraemia.
Introduction A sizeable proportion of the haemodialysis population is transfusion dependent, which can lead to acquired haemosiderosis. Potential clinical consequences of transfusional iron overload are an abnormal liver function, left ventricular dysfunction, endocrine abnormalities, a proximal myopathy, and possibly a particular form of osteomalacia [1-3]. The susceptibility of polytransfused haemodialysis patients to blood-transmitted viral infections has been known since the early years of dialysis [4]. An increased risk of bacterial infections with or without bacteraemia in dialysis patients with transfusional iron overload has only recently been recognised through several retrospective studies [5-7]. The present study is the first to examine in a prospective way the relationship between iron overload, as measured by serum ferritin, and the incidence of bacteraemia in a dialysis population.
Patients and Methods
This prospective study was carried out in one single dialysis unit. All patients treated by maintenance haemodialysis for at least 1 month during a 2-year period from 1 Key words: Bacteraemia; Dialysis; Iron overload October 1986 until 30 September 1988, were enrolled in the study, whether treated in hospital, in a collective autodialysis unit, or at home. For patients entering the study at the start (1 October 1986), the number of units of Correspondence and offprint requests to: Dr J. R. Boelaert, Unit for Renal and Infectious Diseases, Algemeen Ziekenhuis St Jan, B-8000 packed red cells transfused in the past was recorded. Brugge, Belgium. Patients admitted to maintenance haemodialysis after 1
Downloaded from http://ndt.oxfordjournals.org/ at University of Toronto Library on March 23, 2016
Iron Overload in Haemodialysis Patients Increases the Risk of Bacteraemia: A Prospective Study
Iron Overload and Bacteraemia in Dialysis Patients
Results A total of 158 patients entered this study (86 males and 72 females). Their age at inclusion into the study was 1000 ug/1 respectively. The transfusional load in the three patient groups was as follows: in the ferritin < 500 ug/1 group, 8+14 (SD) units of packed red cells over 49 + 47 months;
in the ferritin 500-1000ug/1 group, 22+17 units over 27 + 25 months; and in the ferritin > 1000 ug/1 group, 59 + 32 units over 71 + 50 months. In these 98 patients, the serum ferritin value was significantly correlated (/- = 0.66, i>< 0.001) with the transfusion history, expressed as the number of units of packed red cells transfused per month of dialysis treatment. Over the 2 years of this prospective study the total patient follow-up was 181.3 patient-years. Twenty-nine episodes of bacteraemia were recorded over these 181.3 patient-years (incidence of 0.160 per patient-year): 12 patients developed one bacteraemic episode, seven patients presented two separate episodes and one patient three different episodes. Table 1 shows the yearly incidence of bacteraemia in the three categories of serum ferritin values: 0.1173 for a ferritin 1000 ug/1. As the incidence was similar whether the serum ferritin was < 500 or between 500 and 1000 ug/1, the results of both follow-ups were pooled. The bacteraemic incidence was 2.92 times higher in the ferritin > 1000 than in the ferritin ^ 1000 ug/1 group ( P ^ 0.005). Table 1 shows the incidence of bacteraemia after stratification according to the patient's age (< 65 or ^ 65 years) and according to the treatment duration on maintenance haemodialysis at entry (3 years). An increased rate of bacteraemia was consistently found in the ferritin > 1000 ug/1 group, independently of the age at entry and of the duration of dialysis treatment (P^. 0.005 with the Mantel-Haenszel test). Table 2 lists the 29 episodes of bacteraemia, the microorganism isolated from the blood, the most likely portal of entry of the infection, the patient's renal disease, the last serum ferritin value (^ 1000 or > 1000 ug/1) prior to the bacteraemia and the outcome after the bacteraemic episode. Staph. aureus was isolated in 15, another Gram-positive micro-organism in six, and a Gramnegative bacterium in eight bacteraemic episodes. Seven of the 15 Staph. aureus bacteraemias originated in an infected internal fistula. One of these episodes was complicated by endocarditis, requiring the insertion of a prosthetic valve. Three bacteraemias were related to infection around a central venous dialysis catheter, the cumulative follow-up with a central venous catheter being 6 patient-years; only one of these central venous catheter-related bacteraemias occurred in the ferritin > 1000 ug/1 group. One single bacteraemia occurred during treatment with desferrioxamine. Patients with diabetes mellitus accounted for a total follow-up of 20.2 years (19.4 patient-years in the ferritin jglOOO ug/1 group with three bacteraemias, and 0.75 patient-years in the ferritin > 1000 ug/1 group with two bacteraemias). Limiting the study to non-diabetic patients gives a yearly incidence of bacteraemia of 0.1105 for a ferritin ^ 1000 ug/1 and of 0.2898 for a ferritin > 1000 ug/1.
Downloaded from http://ndt.oxfordjournals.org/ at University of Toronto Library on March 23, 2016
October 1986 were included in this study after 1 month of stabilisation. Blood access was by Cimino-Brescia internal fistula or, when necessary, central venous dialysis catheter. In the latter case, the duration of central venous catheterisation was recorded. Patients were dialysed three times weekly, using dialyser membranes of cuprophane (75%), cellulose acetate (21%), polyacrylonitrile (3%) or polysulphone (1%), with bicarbonate dialysate. Serum ferritin was measured every 3 months by radioimmunoassay (Bio-Rad, Brussels, Belgium; normal values 20300 ug/1). Every 3 months the follow-up of each patient was classified into one of three groups according to the last serum ferritin value: < 500, 500-1000 or > 1000 ug/1. When a patient presented with fever > 38°C, with or without signs of infection, three pairs of blood cultures were obtained using a standard aseptic technique [8]. Blood culture isolates were considered as contaminants if they were Gram-positive and grown from only one or two culture bottles. All bacteraemias not excluded by this criterion were considered as significant. When treatment of the bacteraemic episode required hospitalisation, the total cost of this hospitalisation (including diagnostic work-up and medication) was recorded. At the start of the study, no patient was being treated with recombinant human erythropoietin; during the 2 years of follow-up, 13 patients received this drug. Desferrioxamine was given to only three patients to treat aluminum overload: this drug was given during a total of only 20 patient-months. Sixteen patients received prophylactic nasal mupirocin ointment in order to eradicate nasal Staphylococcus aureus carriage; their follow-up during mupirocin treatment (8.7 patient-years) was not taken into account for the present study [9]. The incidence of bacteraemia per patient-year on dialysis and the outcome after a bacteraemic episode in each of the serum ferritin groups were compared by the Chisquare and the Fisher's exact test respectively. After stratification for the patient's age at entry and the duration of haemodialysis therapy, the incidence of bacteraemia as a function of the serum ferritin was assessed by the Mantel-Haenszel test.
131
J. R. Boelaert et al
132
Table 1. Yearly incidence of bacteraemia in haemodialysis patients according to the serum ferritin value Ferritin groups (ug/1)
=S 65 years > 65 years =S 3 years onHD > 3 years onHD
500-1000
s£1000
>1000
Overall
15/127.92 (0.1173)
2/18.17 (0.1101)
17/146.09 (0.1164) 12/93.25 (0.1266) 5/52.84 (0.0974) 9/90.51 (0.0976) 8/55.58 (0.1484)
12/35.25 (0.3404) 7/19.75 (0.3544) 5/15.50 (0.3225) 7/18.92 (0.3700) 5/16.33 (0.3062)
29/181.34 (0.1599)
The different incidence of bacteraemia between the ferritin > 1000 ug/1 and $ 1000 ug/1 groups is significant (P< 0.005), also after stratification. A, incidence/patient-years; B, annual incidence
Table 2. Details of the episodes of bacteraemia Patient
Ferritin (ug/1) =£1000
l.CIN
+
2. Cong
+ +
3. Diab 4. PK 5. Diab 6. PK 7. CGN 8. CGN 9.?
+ + + + + + + +
Micro-organism isolatpd
Origin of hacteraemia
Outcome
Staph. aureus E. coli Staph. aureus
AV fistula Urinary Central venous catheter AV fistula Pulmonary Urinary ? AV fistula AV fistula 7 Urinary AV fistula Dog bite AV fistula (-(-endocarditis) Central venous catheter Central venous catheter Pulmonary 7 Pulmonary AV fistula ? 7 7 7 7 Peridiverticulitis Tracheostomy wound 7 Cutaneous
R R R
>1000
+ +
+
Staph. aureus Strept. pneumoniae E. coli Staph. aureus Staph. aureus Staph. aureus Strept. pneumoniae E. coli Staph. aureus Pasteurella multocida Staph. aureus Acinetobacter
10. CGN
-r
Siaph. aureus
11. CGN 12. Diab 13. CIN 14. CIN
+ + +
Strept. pneumoniae Slaph. aureus Strept. pneumoniae Staph. aureus Slaph. aureus Pseud, aeruginosa E. coli Staph. aureus Staph. aureus E. coli En terococcus faecalis Strept. viridans Staph. aureus
15. CGN 16. CGN 17.? 18. CGN 19. PK 20. CGN
+
+ + + +
+ + +
R R R R R R R R R R R R
R R R R R R R D
R R R
D R D
CIN, chronic interstitial nephritis; Cong, congenital uropathy; Diab, diabetic glomerulosclerosis; PK, polycystic disease; CGN, chronic glomerulonephritis; ?, unknown. R, recovered; D, died
Downloaded from http://ndt.oxfordjournals.org/ at University of Toronto Library on March 23, 2016
A B A B A B A B A B
Totals
1000 ug/1. However, by the Fisher's exact test, the outcome was not significantly different, whether the ferritin was sj 1000 or > 1000 ug/1. The average cost for the treatment of a bacteraemic episode was approximately two-fold higher in the ferritin > 1000 than in the ^ 1000 ug/1 group (9848 versus 4707 US dollars). However, the higher treatment cost in the high-ferritin class was due to the fact that one Staph. aureus bacteraemia was complicated by endocarditis, which required the insertion of a prosthetic valve and a prolonged hospitalisation.
The prevalence of transfusional iron overload in haemodialysis patients has been assessed by several authors. Crowley et al reported that 15% of the haemodialysis population has received cumulatively more than 10 units of packed red cells [10]. Serum ferritin is generally accepted to be the most accurate simple predictor available of the iron status in dialysis patients [11]. This is confirmed in the present study, where the serum ferritin was significantly related to the transfusional load (/- = 0.66, P < 0.001). The proportion of haemodialysis patients having a high serum ferritin seems to be very variable according to the studies: Rotellar et al [12] reported only 9.5% of the population having a ferritin > 1000 ug/1, whereas Hakim et al [13] reported that as many as 41% of their dialysis population have a ferritin > 2000 ug/1. At the onset of our prospective study, 14 of 98 of the dialysis patients (14.3%) had a serum ferritin > 1000 ug/1. The incidence of bacteraemia in our haemodialysis population over 2 years was 0.160 per patient-year. This corresponds to a yearly incidence between 0.094 and 0.183 previously reported [5-7,14,15]. As in other studies, Staph. aureus was the micro-organism most often responsible for bacteraemia. Although dialysis patients having a central venous dialysis catheter are particularly at risk for Staph. aureus bacteraemia, only two of the 15 Staph. aureus bacteraemias in our study originated at this site. The preponderance of Gram-positive over Gram-negative micro-organisms existed both in the ferritin > 1000 and in the ferritin ^ 1000 ug/1 groups. The present prospective study confirms the results of three previous retrospective studies, indicating that patients with an increased ferritin are at greater risk of bacteraemia [5-7]. Indeed, the incidence of bacteraemia in the ferritin > 1000 ug/1 group was 2.92-fold higher than in the ferritin ^1000 ug/1 group, this difference being significant ( P ^ 0.005). Whereas Tielemans et al found a higher incidence of bacteraemia as soon as the
serum ferritin is > 500 ug/1, the present study disclosed a higher incidence of bacteraemia only for ferritin values > 1000 ug/1, the reason for this difference being unclear [6]. As desferrioxamine was used during only 20 patientmonths, the effect of the drug upon the incidence of bacteraemia cannot be assessed in the present study. Other studies have shown that desferrioxamine does not influence the rate of bacteraemia in dialysis patients [5,16], although the use of this drug in dialysis patients is clearly associated with an increased risk of mucomycosis [17]. Neither the existence of diabetes mellitus in some patients nor the sporadic use of central venous catheters could explain the greater incidence of bacteraemia in the ferritin > 1000 ug/1 group. Some micro-organisms, such as Yersinia enterocolitica and Listeria monocytogenes, are considered to be particularly associated with iron overload. They can cause bacteraemia both in patients with primary haemochromatosis and in iron-overloaded dialysis patients [18,19]. However, in the present study none of these two micro-organisms was isolated. Although the present study discloses a significant association between iron overload in dialysis and an increased incidence of bacteraemia, it was not designed to address the question on a causal relationship between iron overload and an increased susceptibility to infection. This question has been mainly addressed by experimental infections, as reviewed by Weinberg [20]. Two main mechanisms are invoked. The first is the enhancement of bacterial growth in an iron-enriched medium [20]; the second is the decrease in host defence, particularly the function of polymorphonuclear granulocytes, when haemosiderosis is present. Waterlot et al reported that the ability of neutrophils from dialysis patients to phagocytose bakers' yeast is inversely related to the logarithm of the serum ferritin in these patients [21]. We compared the ability of neutrophils to phagocytose Yersinia enterocolitica in dialysis patients with or without iron overload and in healthy individuals, and found that neutrophils from the dialysis patients without iron overload have a normal phagocytosis, whereas neutrophils from iron-overloaded dialysis patients have a severe depression in their ability to phagocytose Yersinia [22]. They have also a killing defect toward this microorganism [22]. In a subsequent study, the administration of recombinant human erythropoietin to iron-overloaded dialysis patients was shown to partially revert these defects in neutrophil function, in parallel with a decrease in ironoverload [23] Treatment of iron-overloaded dialysis patients with erythropoietin is thus expected not only to decrease signs of haemosiderosis, but also to decrease the incidence of bacteraemia and the cost for its treatment. This needs to be documented in the near future.
Downloaded from http://ndt.oxfordjournals.org/ at University of Toronto Library on March 23, 2016
Discussion
133
J. R. Boelaert et al
134
References
Receivedfor publication 20.5.89 Accepted in revisedform 21.8.89
Downloaded from http://ndt.oxfordjournals.org/ at University of Toronto Library on March 23, 2016
1. Schafer AI, Cheron RG, Dluhy R et al. Clinical consequences of acquired transfusional iron overload in adults. NEnglJ Med 1981; 304:319-324 2. Bregman H, Winchester JF, Knepshield JH, Gelfand MC, Manz HJ, Schreiner GE. Iron-overload-associated myopathy in patients on maintenance haemodialysis: a histocompatibility-linked disorder. Lancet 1980; 2: 882-885 3. Van de Vijver FL, Visser WJ, D'Haese PC, De Broe ME. Iron overload and bone disease in chronic dialysis patients. Contrib Nephrol 1988; 64: 134-143 4. Goldman M, Vanherweghem J-L. Multiple blood transfusions and iron overload in patients receiving haemodialysis. Nephrol Dial Transplant 1987; 2: 205-207 5. Seifert A, Van Herrath D, Schaefer K. Iron overload, but not treatment with desferrioxamine favours the development of septicaemia in patients on maintenance haemodialysis. Q J Med 1987; 248: 1015-1024 6. Tielemans C, Lenclud C, Collaert F, Wens R, Dratwa M. Iron status and infections in haemodialysis. Nephrol Dial Transplant 1987; 2: 441 (Abstract) 7. Boelaert J, Schurgers M, Daneels R, Gordts B, Van Landuyt HW. Bacteraemias by Yersinia and other mircroorganisms in haemodialysis patients: a role for iron-overload. Third European Congress of Clinical Microbiology, Den Haag, 1987. Abstract 531 8. Aronson MD, Bor DH. Blood cultures. Ann Intern Med 1987; 106: 246-253 9. Boelaert JR, De Smedt RA, Debaere YA et al. The influence of calcium mupirocin nasal ointment on the incidence of Staphylococcus aureus infections in haemodialysis patients. Nephrol Dial Transplant 1989; 4: 278-281 10. Crowley JP, Nealey TA, Metzger J, Pono L, Chazan JA. Transfusion and long-term hemodialysis. Arch Intern Med 1987; 147: 1925-1928
11. Van de Vijver FL, Vanheule AA, Majelyne WM et al. Serum ferritin as a guide for iron stores in chronic hemodialysis patients. Kidney Int 1984; 24: 451-458 (Letter) 12. Rotellar C, Borgiasz S, Winchester JF, Rotellar E. Am J Kidney Dis 1988; II: 523 (Letter) 13. Hakim RM, Stivelman JC, Schulman G et al. Iron overload and mobilization in long-term hemodialysis patients. Am J Kidney Dis 1987; 10:293-299 14. Dobkin JF, Miller MH, Steigbigel NH. Septicemia in patients on chronic hemodialysis. Ann Intern Med 1978; 88: 28-33 15. Keane WF, Stapiro FL, Raji L. Incidence and type of infections occurring in 445 chronic hemodialysis patients. Trans Am Soc Artif Intern Organs 1977; 23: 41-46 16. Tielemans C, Boelaert J, Vergauwe P et al. Deferoxamine does not increase the risk for bacteremia in hemodialysis patients. Nephron, 1989; 53: 276-277 17. Boelaert JR, Van Roost GE, Vergauwe PL, Verbanck JJ, De Vroey Ch, Segaert MF. The role of desferrioxamine in dialysis-associated mucormycosis: report of three cases and review of the literature. Clin Nephrol 1988; 29: 261-266 18. Boelaert JR, Van Landuyt HW, Valcke YJ et al. The role of iron overload in Yersinia enterocolitica and Yersinia pseudotuberculosis bacteremia in hemodialysis patients. J Infect Dis 1987; 156: 384-387 19. Mossey RT, Sondheimer J. Listerosis in patients with long-term hemodialysis and transfusional iron-overload. Am J Med 1985; 79: 397-400 20. Weinberg ED. Iron and infection. Microbiol Rev 1978; 42: 45-66 21. Waterlot Y, Cantinieaux B, Hariga-Muller C, De MaertelaereLaurent E, Vanherweghem JL, Fondu P. Impaired phagocytic activity of neutrophils in patients receiving haemodialysis: the critical role of iron overload. BrMedJ 1985; 291: 501-504 22. Cantinieaux B, Boelaert J, Hariga C, Fondu P. Impaired neutrophil defense against Yersinia enterocolitica in patients with iron overload who are undergoing dialysis. J Lab Clin Med 1988; 111: 524-528 23. Boelaert J, Cantinieaux B, Hariga C et al. Recombinant human erythropoietin improves neutrophil function in iron-overloaded dialysis patients. Nephrol Dial Transplant 1988; 3: 493 (Abstract)
Nephrology Dialysis Transplantation
Original Article
J. R. Boelaert1, R. F. Daneels1, M. L. Schurgers1, E. G. Matthys1, B. Z. Gordts2 and H. W. Van Landuyt2 'Unit for Renal and Infectious Diseases and 2Department of Microbiology, Algemeen Ziekenhuis St Jan, Brugge, Belgium
Abstract. The incidence of bacteraemia in relation to the degree of transfusional iron overload was studied prospectively in patients from one haemodialysis unit over a 2-year period, with a total follow-up of 181.3 patient-years in 158 patients. Every 3 months, the patients were classified according to the serum ferritin in one of three groups: 1000 ug/1. Twenty-nine episodes of bacteraemia were recorded over 181.3 patient-years (yearly incidence of 0.160). The yearly incidence of bacteraemia was 0.1173 and 0.1101 for ferritin 1000 ug/1 group, the incidence was 0.3404 (P^. 0.005 versus the ferritin ^ 1000 ug/1 group). After stratification for patient's age (at inclusion in the study) and duration of haemodialysis therapy, the higher incidence of bacteraemia in the ferritin >1000 versus < 1000 ug/1 groups persisted (/>< 0.005). This prospective study confirms previous retrospective studies in showing that acquired transfusional iron overload in haemodialysis is associated with a greater risk of bacteraemia.
Introduction A sizeable proportion of the haemodialysis population is transfusion dependent, which can lead to acquired haemosiderosis. Potential clinical consequences of transfusional iron overload are an abnormal liver function, left ventricular dysfunction, endocrine abnormalities, a proximal myopathy, and possibly a particular form of osteomalacia [1-3]. The susceptibility of polytransfused haemodialysis patients to blood-transmitted viral infections has been known since the early years of dialysis [4]. An increased risk of bacterial infections with or without bacteraemia in dialysis patients with transfusional iron overload has only recently been recognised through several retrospective studies [5-7]. The present study is the first to examine in a prospective way the relationship between iron overload, as measured by serum ferritin, and the incidence of bacteraemia in a dialysis population.
Patients and Methods
This prospective study was carried out in one single dialysis unit. All patients treated by maintenance haemodialysis for at least 1 month during a 2-year period from 1 Key words: Bacteraemia; Dialysis; Iron overload October 1986 until 30 September 1988, were enrolled in the study, whether treated in hospital, in a collective autodialysis unit, or at home. For patients entering the study at the start (1 October 1986), the number of units of Correspondence and offprint requests to: Dr J. R. Boelaert, Unit for Renal and Infectious Diseases, Algemeen Ziekenhuis St Jan, B-8000 packed red cells transfused in the past was recorded. Brugge, Belgium. Patients admitted to maintenance haemodialysis after 1
Downloaded from http://ndt.oxfordjournals.org/ at University of Toronto Library on March 23, 2016
Iron Overload in Haemodialysis Patients Increases the Risk of Bacteraemia: A Prospective Study
Iron Overload and Bacteraemia in Dialysis Patients
Results A total of 158 patients entered this study (86 males and 72 females). Their age at inclusion into the study was 1000 ug/1 respectively. The transfusional load in the three patient groups was as follows: in the ferritin < 500 ug/1 group, 8+14 (SD) units of packed red cells over 49 + 47 months;
in the ferritin 500-1000ug/1 group, 22+17 units over 27 + 25 months; and in the ferritin > 1000 ug/1 group, 59 + 32 units over 71 + 50 months. In these 98 patients, the serum ferritin value was significantly correlated (/- = 0.66, i>< 0.001) with the transfusion history, expressed as the number of units of packed red cells transfused per month of dialysis treatment. Over the 2 years of this prospective study the total patient follow-up was 181.3 patient-years. Twenty-nine episodes of bacteraemia were recorded over these 181.3 patient-years (incidence of 0.160 per patient-year): 12 patients developed one bacteraemic episode, seven patients presented two separate episodes and one patient three different episodes. Table 1 shows the yearly incidence of bacteraemia in the three categories of serum ferritin values: 0.1173 for a ferritin 1000 ug/1. As the incidence was similar whether the serum ferritin was < 500 or between 500 and 1000 ug/1, the results of both follow-ups were pooled. The bacteraemic incidence was 2.92 times higher in the ferritin > 1000 than in the ferritin ^ 1000 ug/1 group ( P ^ 0.005). Table 1 shows the incidence of bacteraemia after stratification according to the patient's age (< 65 or ^ 65 years) and according to the treatment duration on maintenance haemodialysis at entry (3 years). An increased rate of bacteraemia was consistently found in the ferritin > 1000 ug/1 group, independently of the age at entry and of the duration of dialysis treatment (P^. 0.005 with the Mantel-Haenszel test). Table 2 lists the 29 episodes of bacteraemia, the microorganism isolated from the blood, the most likely portal of entry of the infection, the patient's renal disease, the last serum ferritin value (^ 1000 or > 1000 ug/1) prior to the bacteraemia and the outcome after the bacteraemic episode. Staph. aureus was isolated in 15, another Gram-positive micro-organism in six, and a Gramnegative bacterium in eight bacteraemic episodes. Seven of the 15 Staph. aureus bacteraemias originated in an infected internal fistula. One of these episodes was complicated by endocarditis, requiring the insertion of a prosthetic valve. Three bacteraemias were related to infection around a central venous dialysis catheter, the cumulative follow-up with a central venous catheter being 6 patient-years; only one of these central venous catheter-related bacteraemias occurred in the ferritin > 1000 ug/1 group. One single bacteraemia occurred during treatment with desferrioxamine. Patients with diabetes mellitus accounted for a total follow-up of 20.2 years (19.4 patient-years in the ferritin jglOOO ug/1 group with three bacteraemias, and 0.75 patient-years in the ferritin > 1000 ug/1 group with two bacteraemias). Limiting the study to non-diabetic patients gives a yearly incidence of bacteraemia of 0.1105 for a ferritin ^ 1000 ug/1 and of 0.2898 for a ferritin > 1000 ug/1.
Downloaded from http://ndt.oxfordjournals.org/ at University of Toronto Library on March 23, 2016
October 1986 were included in this study after 1 month of stabilisation. Blood access was by Cimino-Brescia internal fistula or, when necessary, central venous dialysis catheter. In the latter case, the duration of central venous catheterisation was recorded. Patients were dialysed three times weekly, using dialyser membranes of cuprophane (75%), cellulose acetate (21%), polyacrylonitrile (3%) or polysulphone (1%), with bicarbonate dialysate. Serum ferritin was measured every 3 months by radioimmunoassay (Bio-Rad, Brussels, Belgium; normal values 20300 ug/1). Every 3 months the follow-up of each patient was classified into one of three groups according to the last serum ferritin value: < 500, 500-1000 or > 1000 ug/1. When a patient presented with fever > 38°C, with or without signs of infection, three pairs of blood cultures were obtained using a standard aseptic technique [8]. Blood culture isolates were considered as contaminants if they were Gram-positive and grown from only one or two culture bottles. All bacteraemias not excluded by this criterion were considered as significant. When treatment of the bacteraemic episode required hospitalisation, the total cost of this hospitalisation (including diagnostic work-up and medication) was recorded. At the start of the study, no patient was being treated with recombinant human erythropoietin; during the 2 years of follow-up, 13 patients received this drug. Desferrioxamine was given to only three patients to treat aluminum overload: this drug was given during a total of only 20 patient-months. Sixteen patients received prophylactic nasal mupirocin ointment in order to eradicate nasal Staphylococcus aureus carriage; their follow-up during mupirocin treatment (8.7 patient-years) was not taken into account for the present study [9]. The incidence of bacteraemia per patient-year on dialysis and the outcome after a bacteraemic episode in each of the serum ferritin groups were compared by the Chisquare and the Fisher's exact test respectively. After stratification for the patient's age at entry and the duration of haemodialysis therapy, the incidence of bacteraemia as a function of the serum ferritin was assessed by the Mantel-Haenszel test.
131
J. R. Boelaert et al
132
Table 1. Yearly incidence of bacteraemia in haemodialysis patients according to the serum ferritin value Ferritin groups (ug/1)
=S 65 years > 65 years =S 3 years onHD > 3 years onHD
500-1000
s£1000
>1000
Overall
15/127.92 (0.1173)
2/18.17 (0.1101)
17/146.09 (0.1164) 12/93.25 (0.1266) 5/52.84 (0.0974) 9/90.51 (0.0976) 8/55.58 (0.1484)
12/35.25 (0.3404) 7/19.75 (0.3544) 5/15.50 (0.3225) 7/18.92 (0.3700) 5/16.33 (0.3062)
29/181.34 (0.1599)
The different incidence of bacteraemia between the ferritin > 1000 ug/1 and $ 1000 ug/1 groups is significant (P< 0.005), also after stratification. A, incidence/patient-years; B, annual incidence
Table 2. Details of the episodes of bacteraemia Patient
Ferritin (ug/1) =£1000
l.CIN
+
2. Cong
+ +
3. Diab 4. PK 5. Diab 6. PK 7. CGN 8. CGN 9.?
+ + + + + + + +
Micro-organism isolatpd
Origin of hacteraemia
Outcome
Staph. aureus E. coli Staph. aureus
AV fistula Urinary Central venous catheter AV fistula Pulmonary Urinary ? AV fistula AV fistula 7 Urinary AV fistula Dog bite AV fistula (-(-endocarditis) Central venous catheter Central venous catheter Pulmonary 7 Pulmonary AV fistula ? 7 7 7 7 Peridiverticulitis Tracheostomy wound 7 Cutaneous
R R R
>1000
+ +
+
Staph. aureus Strept. pneumoniae E. coli Staph. aureus Staph. aureus Staph. aureus Strept. pneumoniae E. coli Staph. aureus Pasteurella multocida Staph. aureus Acinetobacter
10. CGN
-r
Siaph. aureus
11. CGN 12. Diab 13. CIN 14. CIN
+ + +
Strept. pneumoniae Slaph. aureus Strept. pneumoniae Staph. aureus Slaph. aureus Pseud, aeruginosa E. coli Staph. aureus Staph. aureus E. coli En terococcus faecalis Strept. viridans Staph. aureus
15. CGN 16. CGN 17.? 18. CGN 19. PK 20. CGN
+
+ + + +
+ + +
R R R R R R R R R R R R
R R R R R R R D
R R R
D R D
CIN, chronic interstitial nephritis; Cong, congenital uropathy; Diab, diabetic glomerulosclerosis; PK, polycystic disease; CGN, chronic glomerulonephritis; ?, unknown. R, recovered; D, died
Downloaded from http://ndt.oxfordjournals.org/ at University of Toronto Library on March 23, 2016
A B A B A B A B A B
Totals
1000 ug/1. However, by the Fisher's exact test, the outcome was not significantly different, whether the ferritin was sj 1000 or > 1000 ug/1. The average cost for the treatment of a bacteraemic episode was approximately two-fold higher in the ferritin > 1000 than in the ^ 1000 ug/1 group (9848 versus 4707 US dollars). However, the higher treatment cost in the high-ferritin class was due to the fact that one Staph. aureus bacteraemia was complicated by endocarditis, which required the insertion of a prosthetic valve and a prolonged hospitalisation.
The prevalence of transfusional iron overload in haemodialysis patients has been assessed by several authors. Crowley et al reported that 15% of the haemodialysis population has received cumulatively more than 10 units of packed red cells [10]. Serum ferritin is generally accepted to be the most accurate simple predictor available of the iron status in dialysis patients [11]. This is confirmed in the present study, where the serum ferritin was significantly related to the transfusional load (/- = 0.66, P < 0.001). The proportion of haemodialysis patients having a high serum ferritin seems to be very variable according to the studies: Rotellar et al [12] reported only 9.5% of the population having a ferritin > 1000 ug/1, whereas Hakim et al [13] reported that as many as 41% of their dialysis population have a ferritin > 2000 ug/1. At the onset of our prospective study, 14 of 98 of the dialysis patients (14.3%) had a serum ferritin > 1000 ug/1. The incidence of bacteraemia in our haemodialysis population over 2 years was 0.160 per patient-year. This corresponds to a yearly incidence between 0.094 and 0.183 previously reported [5-7,14,15]. As in other studies, Staph. aureus was the micro-organism most often responsible for bacteraemia. Although dialysis patients having a central venous dialysis catheter are particularly at risk for Staph. aureus bacteraemia, only two of the 15 Staph. aureus bacteraemias in our study originated at this site. The preponderance of Gram-positive over Gram-negative micro-organisms existed both in the ferritin > 1000 and in the ferritin ^ 1000 ug/1 groups. The present prospective study confirms the results of three previous retrospective studies, indicating that patients with an increased ferritin are at greater risk of bacteraemia [5-7]. Indeed, the incidence of bacteraemia in the ferritin > 1000 ug/1 group was 2.92-fold higher than in the ferritin ^1000 ug/1 group, this difference being significant ( P ^ 0.005). Whereas Tielemans et al found a higher incidence of bacteraemia as soon as the
serum ferritin is > 500 ug/1, the present study disclosed a higher incidence of bacteraemia only for ferritin values > 1000 ug/1, the reason for this difference being unclear [6]. As desferrioxamine was used during only 20 patientmonths, the effect of the drug upon the incidence of bacteraemia cannot be assessed in the present study. Other studies have shown that desferrioxamine does not influence the rate of bacteraemia in dialysis patients [5,16], although the use of this drug in dialysis patients is clearly associated with an increased risk of mucomycosis [17]. Neither the existence of diabetes mellitus in some patients nor the sporadic use of central venous catheters could explain the greater incidence of bacteraemia in the ferritin > 1000 ug/1 group. Some micro-organisms, such as Yersinia enterocolitica and Listeria monocytogenes, are considered to be particularly associated with iron overload. They can cause bacteraemia both in patients with primary haemochromatosis and in iron-overloaded dialysis patients [18,19]. However, in the present study none of these two micro-organisms was isolated. Although the present study discloses a significant association between iron overload in dialysis and an increased incidence of bacteraemia, it was not designed to address the question on a causal relationship between iron overload and an increased susceptibility to infection. This question has been mainly addressed by experimental infections, as reviewed by Weinberg [20]. Two main mechanisms are invoked. The first is the enhancement of bacterial growth in an iron-enriched medium [20]; the second is the decrease in host defence, particularly the function of polymorphonuclear granulocytes, when haemosiderosis is present. Waterlot et al reported that the ability of neutrophils from dialysis patients to phagocytose bakers' yeast is inversely related to the logarithm of the serum ferritin in these patients [21]. We compared the ability of neutrophils to phagocytose Yersinia enterocolitica in dialysis patients with or without iron overload and in healthy individuals, and found that neutrophils from the dialysis patients without iron overload have a normal phagocytosis, whereas neutrophils from iron-overloaded dialysis patients have a severe depression in their ability to phagocytose Yersinia [22]. They have also a killing defect toward this microorganism [22]. In a subsequent study, the administration of recombinant human erythropoietin to iron-overloaded dialysis patients was shown to partially revert these defects in neutrophil function, in parallel with a decrease in ironoverload [23] Treatment of iron-overloaded dialysis patients with erythropoietin is thus expected not only to decrease signs of haemosiderosis, but also to decrease the incidence of bacteraemia and the cost for its treatment. This needs to be documented in the near future.
Downloaded from http://ndt.oxfordjournals.org/ at University of Toronto Library on March 23, 2016
Discussion
133
J. R. Boelaert et al
134
References
Receivedfor publication 20.5.89 Accepted in revisedform 21.8.89
Downloaded from http://ndt.oxfordjournals.org/ at University of Toronto Library on March 23, 2016
1. Schafer AI, Cheron RG, Dluhy R et al. Clinical consequences of acquired transfusional iron overload in adults. NEnglJ Med 1981; 304:319-324 2. Bregman H, Winchester JF, Knepshield JH, Gelfand MC, Manz HJ, Schreiner GE. Iron-overload-associated myopathy in patients on maintenance haemodialysis: a histocompatibility-linked disorder. Lancet 1980; 2: 882-885 3. Van de Vijver FL, Visser WJ, D'Haese PC, De Broe ME. Iron overload and bone disease in chronic dialysis patients. Contrib Nephrol 1988; 64: 134-143 4. Goldman M, Vanherweghem J-L. Multiple blood transfusions and iron overload in patients receiving haemodialysis. Nephrol Dial Transplant 1987; 2: 205-207 5. Seifert A, Van Herrath D, Schaefer K. Iron overload, but not treatment with desferrioxamine favours the development of septicaemia in patients on maintenance haemodialysis. Q J Med 1987; 248: 1015-1024 6. Tielemans C, Lenclud C, Collaert F, Wens R, Dratwa M. Iron status and infections in haemodialysis. Nephrol Dial Transplant 1987; 2: 441 (Abstract) 7. Boelaert J, Schurgers M, Daneels R, Gordts B, Van Landuyt HW. Bacteraemias by Yersinia and other mircroorganisms in haemodialysis patients: a role for iron-overload. Third European Congress of Clinical Microbiology, Den Haag, 1987. Abstract 531 8. Aronson MD, Bor DH. Blood cultures. Ann Intern Med 1987; 106: 246-253 9. Boelaert JR, De Smedt RA, Debaere YA et al. The influence of calcium mupirocin nasal ointment on the incidence of Staphylococcus aureus infections in haemodialysis patients. Nephrol Dial Transplant 1989; 4: 278-281 10. Crowley JP, Nealey TA, Metzger J, Pono L, Chazan JA. Transfusion and long-term hemodialysis. Arch Intern Med 1987; 147: 1925-1928
11. Van de Vijver FL, Vanheule AA, Majelyne WM et al. Serum ferritin as a guide for iron stores in chronic hemodialysis patients. Kidney Int 1984; 24: 451-458 (Letter) 12. Rotellar C, Borgiasz S, Winchester JF, Rotellar E. Am J Kidney Dis 1988; II: 523 (Letter) 13. Hakim RM, Stivelman JC, Schulman G et al. Iron overload and mobilization in long-term hemodialysis patients. Am J Kidney Dis 1987; 10:293-299 14. Dobkin JF, Miller MH, Steigbigel NH. Septicemia in patients on chronic hemodialysis. Ann Intern Med 1978; 88: 28-33 15. Keane WF, Stapiro FL, Raji L. Incidence and type of infections occurring in 445 chronic hemodialysis patients. Trans Am Soc Artif Intern Organs 1977; 23: 41-46 16. Tielemans C, Boelaert J, Vergauwe P et al. Deferoxamine does not increase the risk for bacteremia in hemodialysis patients. Nephron, 1989; 53: 276-277 17. Boelaert JR, Van Roost GE, Vergauwe PL, Verbanck JJ, De Vroey Ch, Segaert MF. The role of desferrioxamine in dialysis-associated mucormycosis: report of three cases and review of the literature. Clin Nephrol 1988; 29: 261-266 18. Boelaert JR, Van Landuyt HW, Valcke YJ et al. The role of iron overload in Yersinia enterocolitica and Yersinia pseudotuberculosis bacteremia in hemodialysis patients. J Infect Dis 1987; 156: 384-387 19. Mossey RT, Sondheimer J. Listerosis in patients with long-term hemodialysis and transfusional iron-overload. Am J Med 1985; 79: 397-400 20. Weinberg ED. Iron and infection. Microbiol Rev 1978; 42: 45-66 21. Waterlot Y, Cantinieaux B, Hariga-Muller C, De MaertelaereLaurent E, Vanherweghem JL, Fondu P. Impaired phagocytic activity of neutrophils in patients receiving haemodialysis: the critical role of iron overload. BrMedJ 1985; 291: 501-504 22. Cantinieaux B, Boelaert J, Hariga C, Fondu P. Impaired neutrophil defense against Yersinia enterocolitica in patients with iron overload who are undergoing dialysis. J Lab Clin Med 1988; 111: 524-528 23. Boelaert J, Cantinieaux B, Hariga C et al. Recombinant human erythropoietin improves neutrophil function in iron-overloaded dialysis patients. Nephrol Dial Transplant 1988; 3: 493 (Abstract)
