1008

Intravenous Colistin as Therapy for Nosocomial Infections Caused by Multidrug-

Resistant Pseudomonas aeruginosa and Acinetobacter baumannii
Anna S. Levin, Antonio A. Barone, Juliana Penço, From the Hospital Infection Control Department, Hospital das ClıB nicas,
Marcio V. Santos, Ivan S. Marinho, Erico A. G. Arruda, Faculdade de Medicina, University of São Paulo, São Paulo, Brazil
Edison I. Manrique, and Silvia F. Costa

Sixty nosocomial infections caused by Pseudomonas aeruginosa and Acinetobacter baumannii

resistant to aminoglycosides, cephalosporins, quinolones, penicillins, monobactams, and imipenem
were treated with colistin (one patient had two infections that are included as two different cases).
The infections were pneumonia (33% of patients), urinary tract infection (20%), primary bloodstream

Downloaded from https://academic.oup.com/cid/article-abstract/28/5/1008/324319 by guest on 14 March 2019

infection (15%), central nervous system infection (8%), peritonitis (7%), catheter-related infection
(7%), and otitis media (2%). A good outcome occurred for 35 patients (58%), and three patients
died within the first 48 hours of treatment. The poorest results were observed in cases of pneumonia:
only five (25%) of 20 had a good outcome. A good outcome occurred for four of five patients with
central nervous system infections, although no intrathecal treatment was given. The main adverse
effect of treatment was renal failure; 27% of patients with initially normal renal function had renal
failure, and renal function worsened in 58% of patients with abnormal baseline creatinine levels.
Colistin may be a good therapeutic option for the treatment of severe infections caused by multidrug-
resistant P. aeruginosa and A. baumannii.

The emergence of multidrug-resistant gram-negative organ-
isms causing nosocomial infections is a growing problem
worldwide [1, 2]. In our hospital, Acinetobacter baumannii and
Pseudomonas aeruginosa resistant to all commercially avail-
able antimicrobial drugs have been important causes of infec-
tions acquired mainly in intensive care units [3, 4]. Hospital
das ClıB nicas is a 2,200-bed tertiary care hospital attached to
the University of São Paulo (São Paulo, Brazil). Infection due
to multidrug-resistant A. baumannii has been a problem since
1993 when an outbreak occurred. All isolates were one clone,
and the outbreak was initially controlled; however, clusters of
cases have been occurring since then [3]. Multidrug-resistant
P. aeruginosa infections have been investigated in our hospital;
four different typing methods have revealed that these infec-
tions are caused by multiple clones (authors’ unpublished data).
Colistin, a polymyxin, was used from the 1960s to the early
1980s. Because of toxicity considerations (mainly nephrotoxic-
ity, neuromuscular blockade, and neurotoxicity), its systemic
use has been all but abandoned [5, 6]. The mechanism of action
of colistin is not very clear; however, it is believed that its
bactericidal activity is due to a detergent effect on the cell
membrane [6, 7]. In vitro colistin has shown excellent activity
against a variety of gram-negative rods including those resistant
to other classes of antimicrobials (penicillins, cephalosporins,
quinolones, aminoglycosides, and carbapenems), although very
Received 12 June 1998; revised 23 November 1998.
Reprints or correspondence: Dr. Anna S. Levin, Rua Harmonia, 564/52, São
Paulo-SP 05435-000, Brazil (nivel@usp.br).
Clinical Infectious Diseases 1999;28:1008–11
q 1999 by the Infectious Diseases Society of America. All rights reserved.
1058–4838/99/2805–0008$03.00
little activity has been demonstrated against Serratia species,
Providencia species, and Proteus mirabilis [5, 6].
All the multidrug-resistant isolates of P. aeruginosa and
A. baumannii in our hospital were susceptible in vitro to colis-
tin. The objective of this study was to evaluate the use of
colistin in the treatment of infections caused by these multi-
drug-resistant microorganisms.
Methods
Patients with infections caused by multidrug-resistant P. aer-
uginosa or A. baumannii who were hospitalized during the
period from January 1993 to December 1994 were treated with
intravenous colistin (colistimethate sodium, Parke-Davis, Mor-
ris Plains, NJ; or colistin sodium methane sulfonate, Bellon,
Rhône-Poulenc Rorer, Neuilly sur Seine, France).
All infections were diagnosed according to the criteria of
the Centers for Disease Control and Prevention [8]. On the
basis of their etiologies, the cases were considered confirmed
or probable. A confirmed case was defined as a nosocomial
infection with a culture of a specimen obtained from a usually
sterile site (e.g., blood or CSF) that was positive for P. aerugi-
nosa or A. baumannii. A probable case was defined as an
infection whose etiology could not be confirmed because the
positive culture was of a specimen from a usually nonsterile
site for which there was no evidence of another cause.
Identification of P. aeruginosa and A. baumannii was per-
formed by using classic biochemical methods [9, 10]. Suscepti-
bility testing was done according to an automated method
(bioMérieux Vitek, Hazelwood, MO); susceptibility to various
antimicrobials (gram-negative susceptibility cards PA and GD)
was analyzed. Susceptibility to colistin was tested by means
of the disk diffusion method with use of a 10-mg colistin disk

/ 9c65$$my03 04-06-99 07:38:07 cida UC: CID

CID 1999;28 (May) Colistin Therapy for Nosocomial Infections
(Oxoid, Basingstoke, Hants, England), and isolates were con-
sidered susceptible if the inhibition zone was §11 mm.
Only infections for which there was no other therapeutic
option were treated with colistin. Patients were evaluated at
the beginning of treatment for the following data: age, gender,
identity of infectious agent, and severity of clinical condition
according to the APACHE (Acute Physiological and Chronic
Health Evaluation) II scoring system [11]. Patients were fol-
lowed up until the end of the treatment for outcome that was
considered improved or worsened based on clinical criteria,
the evaluation of the attending physician, and adverse events.
The patients were treated with 2.5 to 5.0 mg of colistin/kg
daily up to a maximum dose of 300 mg, which was divided
in two or three doses intravenously. When the patients pre-
sented with renal failure, the daily dose was adjusted as follows:
serum creatinine level from 1.3 to 1.5 mg/dL, daily dose of
2.5 to 5.0 mg/kg; 1.6 to 2.5 mg/dL, 2.5 mg/kg; and ú2.5
mg/dL, 1.0 to 1.5 mg/kg.
The isolates were resistant to the following antimicrobials
on the basis of the MICs: ticarcillin and piperacillin (MIC,
ú64 mg/mL), ceftazidime (MIC, §64 mg/mL), aztreonam
(MIC, §32 mg/mL), amikacin (MIC, §32 mg/mL), gentamicin
(MIC, §8 mg/mL), ciprofloxacin (MIC, §4 mg/mL), and imi-
penem (MIC, ú8 mg/mL). The MIC breakpoints for suscepti-
bility according to the National Committee for Clinical Labora-
tory Standards [12] are as follows: 64 mg of ticarcillin/mL or
64 mg of piperacillin/mL, P. aeruginosa; 16 mg of piperacillin/
mL or 16 mg of ticarcillin/mL, A. baumannii; 8 mg of ceftazi-
dime/mL, both microorganisms; 8 mg of aztreonam/mL, both
microorganisms; 16 mg of amikacin/mL, both microorganisms;
4 mg of gentamicin/mL, both microorganisms; 1 mg of ci-
profloxacin/mL, both microorganisms; and 4 mg of imipenem/
mL, both microorganisms.
Results
Sixty infections in 59 patients were treated with colistin (one
patient had two infections that are included as two different
cases); 21 (35%) were caused by P. aeruginosa, and 39 (65%)
were caused by A. baumannii. The mean age { SD of the
patients was 42.1 { 21.4 years (range, 2 – 85 years), and 39
(65%) were male. The specimens from which the microorgan-
isms were isolated are listed in table 1.
The etiology of 42 infections (70%) was confirmed, and 18
cases (30%) were probable.
The underlying diseases in the 59 patients are listed in table
2. The mean APACHE II score { SD was 13.1 { 7.0 (range,
2 – 34). Thirty-one infections (52%) occurred in intensive care
units, 8 (13%) in transplant units, and 21 (35%) in medical or
surgical wards. The mean hospital stay before infection { SD
was 38.4 { 37.1 days (range, 0 – 237 days). Fifty-six infections
(93%) had been previously treated with other antimicrobials:
imipenem, 30 (54%) of the treated cases; aminoglycosides, 20
/ 9c65$$my03 04-06-99 07:38:07
1009
(36%); ceftazidime, 16 (29%); quinolones, 16 (29%); and third-
generation cephalosporins, 15 (27%).
A good outcome occurred in 35 (58%) of the cases. Of the
confirmed cases, 28 (67%) had a good outcome, and of the
probable cases, seven (39%) had a good outcome. The infec-
tions and their outcomes are shown in table 1. In the cases
with a good outcome, fever lasted for a mean duration { SD
of 5.5 { 5.8 days (range, 0 – 21 days). Follow-up cultures of
specimens from the original positive site were performed in
29 cases; 27 (93%) of these cultures were negative. Death
occurred in 22 cases (37%); three of the patients died within
Downloaded from https://academic.oup.com/cid/article-abstract/28/5/1008/324319 by guest on 14 March 2019
the first 48 hours of treatment.
The mean daily dose of colistin { SD was 152.8 { 62.8
mg (range, 60 – 300 mg); the mean duration of treatment { SD
was 12.6 { 6.8 days (range, 2 – 34 days). The mean duration
of treatment { SD for the surviving patients was 14.0 { 5.1
days (range, 5 – 25 days).
The mean baseline serum creatinine level { SD was 1.5 {
1.5 mg/dL (range, 0.1 – 9.1 mg/dL), and 19 (32%) of the pa-
tients presented with renal failure defined as a creatinine level
of ú1.5 mg/dL or a urea level of ú50 mg/dL. Eleven (27%)
of the 41 patients with normal renal function at the initial
evaluation presented with worsening renal function during
treatment (mean increase in creatinine level { SD, 0.9 { 0.6
mg/dL; range, 0.3 – 2.4 mg/dL). Follow-up was possible for six
of these 11 patients, and the creatinine levels became normal
in three within 1 month after the end of treatment. Of 19
patients with abnormal baseline creatinine levels, 11 (58%) had
worsening renal function during treatment (mean increase in
creatinine level { SD, 1.5 { 1.4 mg/dL; range, 0.1 – 4.0
mg/dL). Nephrotoxicity did not cause discontinuation of treat-
ment. No neuromuscular disorders were observed.
Discussion
The emergence of multidrug-resistant P. aeruginosa and
A. baumannii causing nosocomial infections poses a very seri-
ous therapeutic problem in Brazil as there are no commercially
available alternatives for treatment. Until the early 1980s, colis-
tin was used for treating infections caused by gram-negative
rods. The use of colistin was dropped when second- and third-
generation cephalosporins became available, mainly because of
toxicity [5]. Ticarcillin was commercialized only in the 1990s.
Colistin is not available in Brazil, although it is an approved
drug (thus allowing importation from other countries).
Colistin was reported as a new agent in 1952 and was shown
to be polymyxin E, belonging to the polymyxins (a polypeptide
antibiotic group discovered in 1947 – 1948). Polymyxin E is
sulfomethylated, obliterating its cationic nature and rendering
it ineffective as an antibiotic. The sulfomethyl polymyxin must
be hydrolyzed to release the active free base, and this release
occurs at body temperature and at physiological pH in aqueous
systems. The sulfomethylated form of the drug is called colisti-
methate sodium or colistin sodium methane sulfonate [13]. The
cida UC: CID
1010 Levin et al. CID 1999;28 (May)

Table 1. Nosocomial infections caused by multidrug-resistant Pseudomonas aeruginosa and Acineto-

bacter baumannii: isolates and outcomes.

No. (%) No. (%) with No. of isolates

of good
Infection infections outcome Specimen(s) (no.) P. aeruginosa A. baumannii

Pneumonia 20 (33) 5 (25) Tracheal secretion (14), pleural 6 14

fluid (3), blood (2), BAL
fluid (1)
Urinary tract 12 (20) 10 (83) Urine (12) 4 8
infection
Primary bloodstream 9 (15) 7 (78) Blood (9) 7 2

Downloaded from https://academic.oup.com/cid/article-abstract/28/5/1008/324319 by guest on 14 March 2019

infection
CNS infection 5 (8) 4 (80) CSF (5) 0 5
Surgical site 5 (8) 3 (60) Surgical site specimen (5) 1 4
infection
Peritonitis 4 (7) 2 (50) Peritoneal fluid (4) 1 3
Catheter-related 4 (7) 3 (75) Central venous catheter (4) 1 3
infection
Otitis media 1 (2) 1 (100) Middle ear fluid (1) 1 0
Total 60 (100) 35 (58) 21 39

NOTE. BAL Å bronchoalveolar lavage.

antibacterial action of colistin is poorly understood, but the other available drugs as well as infections due to other gram-
bacterial cell membrane is disrupted by the binding of the drug negative bacilli. These infections included septicemia and uri-
to phospholipids, leading to the death of the organism. The nary tract, wound, and respiratory tract infections [16 – 20].
drug binds to acidic phospholipids in mammalian tissue, where There are also studies reporting good results of colistin treat-
it has no antibacterial activity [7] but may account for the ment of children and newborns [21, 22]. In our study, colistin
persistent detectable levels of the free form that might be slowly was used only in cases in which there were no therapeutic
released from binding sites. Colistin has no effect against gram- alternatives. The patients treated were mostly from intensive
positive bacteria, has an insignificant antifungal effect, and is care units, which are mostly affected by multidrug-resistant
effective against gram-negative bacilli (including P. aerugi- organisms, and were severely ill as can be seen by their under-
nosa, Escherichia coli, Klebsiella pneumoniae, Enterobacter lying conditions and APACHE II scores. A good outcome
species, Salmonella species, Shigella species, Vibrio species, occurred in 35 (58%) of 60 cases. Three patients died within
Pasteurella species, Haemophilus species, and Bordetella spe- the first 48 hours of treatment, and these patients’ treatments
cies); Proteus species, Providencia species, and most Serratia cannot really be considered as failures.
isolates are resistant to colistin [14, 15]. The infection associated with the poorest results was pneu-
In the past (1960s and 1970s), colistimethate sodium was monia, for which 15 of 20 treatments failed. This occurrence
used to treat infections caused by P. aeruginosa resistant to cannot be explained by the fact that in most of these cases
the etiologies were probable and not confirmed, because four
of five confirmed cases had a bad outcome. Results were good
for a variety of different infections, and surprisingly, four of
Table 2. Underlying diseases in 59 patients with infections caused
five cases of CNS infection (meningitis and ventriculitis) had
by multidrug-resistant microorganisms who were treated with colistin.
a good outcome, although no intrathecal therapy was given.
Underlying disease No. (%) of patients Colistin has been reported to have poor penetration through
the blood-brain barrier [6], but our study and other investiga-
Neurological disorder 14 (24) tors [22] who reported good results of colistin treatment sug-
Abdominal surgery 10 (17) gest that further studies are needed to clarify this issue. A
Hematologic malignancy 7 (12)
possible explanation is the improvement of penetration with
Transplantation (bone marrow, kidney, or liver) 7 (12)
Burn 6 (11) inflammation.
Tetanus 5 (8) Toxicity is an important concern with colistin, and it has
Hepatic cirrhosis 5 (8) never been a first choice for treatment of infections due to gram-
Trauma 3 (5) negative bacteria because of this concern, although toxicity was
Chagas’ disease 2 (3)
not recognized at first [13]. Nephrotoxicity is the most im-
Total 59 (100)
portant adverse effect. In normal subjects, transient renal im-

/ 9c65$$my03 04-06-99 07:38:07 cida UC: CID

CID 1999;28 (May) Colistin Therapy for Nosocomial Infections
pairment may occur during therapy, but in patients with pre-
existing renal disease, severe renal impairment with residual
damage occurs more frequently. Doses for patients with renal
insufficiency must be adjusted, as colistin is excreted princi-
pally by the kidneys and elevated blood levels of the drug may
further impair renal function [23 – 25]. In our study, nephrotox-
icity was observed in 22 patients (37%), occurring more fre-
quently in patients with preexisting alterations of renal func-
tion. No treatments were stopped because of adverse effects of
the drug. Neurotoxicity and neuromuscular blockade have also
been reported with the use of colistin [25, 26], but these effects
were not observed in our study; however, the fact that many
patients were receiving support with mechanical ventilation
and were sedated made evaluation difficult.
This study shows that colistin may be a good therapeutic
option in the treatment of severe infections caused by multi-
drug-resistant P. aeruginosa and A. baumannii.
References
1. Bergogne-Bérézin E, Towner KJ. Acinetobacter spp. nosocomial patho-
gens: microbiological, clinical, and epidemiological features. Clin Mi-
crobiol Rev 1996; 9:148 – 65.
2. Livermore DM. Are all b-lactams created equal? Scand J Infect Dis 1996;
101:33 – 43.
3. Levin AS, Mendes CMF, Sinto SI, et al. An outbreak of multiresistant
Acinetobacter baumannii in a university hospital in São Paulo, Brazil.
Infect Control Hosp Epidemiol 1996; 17:366 – 8.
4. Sader HS, Pignatari AC, Leme JL, et al. Epidemiologic typing of multiple
drug-resistant Pseudomonas aeruginosa isolated from an outbreak in
an intensive care unit. Diagn Microbiol Infect Dis 1993; 17:13 – 8.
5. Catchpole CR, Andrews JM, Brenwald N, Wise R. A reassessment of
the in-vitro activity of colistin sulphomethate sodium. J Antimicrob
Chemother 1997; 39:255 – 60.
6. Horton J, Pankey GA. Polymyxin B, colistin, and sodium colistimethate.
Med Clin North Am 1982; 66:135 – 42.
7. Kunin CM, Bugg A. Binding of polymyxin antibiotics to tissues: the major
determinant of distribution and persistence in the body. J Infect Dis
1971; 124:394 – 400.
/ 9c65$$my03 04-06-99 07:38:07
1011
8. Garner JS, Jarvis WR, Emori TG, Horn TC, Hughes JM. CDC definitions
for nosocomial infections, 1988. Am J Infect Control 1988; 16:128 – 40.
9. Gilligan PH. Pseudomonas and Burkholderia. In: Murray PR, Baron EJ,
Pfaller MA, Tenover FC, Yolken RH, eds. Manual of clinical microbiol-
ogy. 6th ed. Washington, DC: ASM Press, 1995:509 – 19.
10. Gravenitz A. Acinetobacter, Alcaligenes, Moraxella and other nonfermen-
tative gram-negative bacteria. In: Murray PR, Baron EJ, Pfaller MA,
Tenover FC, Yolken RH, eds. Manual of clinical microbiology. 6th ed.
Washington, DC: ASM Press, 1995:520 – 32.
11. Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II: a
severity of disease classification system. Crit Care Med 1985; 13:
818 – 29.
12. National Committee for Clinical Laboratory Standards. Performance stan-
dards for antimicrobial susceptibility testing. Vol 14. NCCLS document
Downloaded from https://academic.oup.com/cid/article-abstract/28/5/1008/324319 by guest on 14 March 2019
M2-M5 (M100-S5). Villanova, Pennsylvania: National Committee for
Clinical Laboratory Standards, 1994.
13. Barnett M, Bushby SRM, Wilkinson S. Sodium sulphomethyl derivatives
of polymyxins. Br J Pharmacol 1964; 23:552 – 74.
14. Hoeprich PD. The polymyxins. Med Clin North Am 1970; 54:1257 – 65.
15. Goodwin NJ. Colistin and sodium colistimethate. Med Clin North Am
1970; 54:1267 – 76.
16. Petersdorf RG, Plorde JJ. Colistin — a reappraisal. JAMA 1963; 183:
123 – 5.
17. Yow EM, Tan E, Shane L, Schonfeld S, Abu-Nassar H. Colistin (Coly-
Mycin) in resistant bacterial infections. Arch Intern Med 1961; 108:
64 – 70.
18. Kirby WMM, Roberts CE Jr. Colistin. JAMA 1961; 177:112.
19. Edgar WM, Dickinson KM. A trial of colistin methane sulphonate in
urinary infection with Pseudomonas pyocyanea. Lancet 1962; 2:
739 – 40.
20. Fekety FR Jr, Norman PS, Cluff LE. The treatment of gram-negative
bacillary infections with colistin. Ann Intern Med 1962; 57:214 – 29.
21. Walker SH, Patron LR. Colistin therapy in infants and children. Am J Dis
Child 1965; 109:204 – 11.
22. Marsden HB, Hyde WA. Colistin methane sulphonate in childhood infec-
tions. Lancet 1962; 2:740 – 2.
23. Goodwin NJ, Friedman EA. The effects of renal impairment, peritoneal
dialysis, and hemodialysis on serum sodium colistimethate levels. Ann
Intern Med 1968; 68:984 – 94.
24. MacKay DN, Kaye D. Serum concentrations of colistin in patients with
normal and impaired renal function. N Engl J Med 1964; 270:394 – 7.
25. Koch-Weser J, Sidel VW, Federman EB, Kanarek P, Finer DC, Eaton AE.
Adverse effects of sodium colistimethate. Ann Intern Med 1970; 72:
857 – 68.
26. Wolinsky E, Hines JD. Neurotoxic and nephrotoxic effects of colistin in
patients with renal disease. N Engl J Med 1962; 266:759 – 62.
cida UC: CID