- Top of page
- Patients and methods
Following a 2-year study, the combination of oral ciprofloxacin and colistin has been used continuously for 10 years without the emergence of resistance. During a 2-year period (1987–1989), we compared ciprofloxacin + colistin (CIP + COL) with neomycin + colistin (NEO + COL) in a randomized trial – combinations chosen because of the potential for prophylaxis of Gram-negative infection by ciprofloxacin, with colistin given to reduce the risk of emergence of resistance. Sixty-four patients with similar demographics in each arm were evaluable for efficacy analysis. Patients on CIP + COL had a significantly lower proportion of neutropenic days with fever (P < 0·001) and neutropenic days on intravenous antibiotics (P < 0·001) than patients on NEO + COL. A total of 54 (15 bacteriologically documented) pyrexial episodes occurred in patients on CIP + COL and 77 (41 bacteriologically documented) in patients on NEO + COL. Only two Gram-negative bacterial infections occurred in the CIP + COL arm compared with 16 in the NEO + COL arm. No Staphylococcus aureus infections occurred in the CIP + COL group compared with 10 in the other patients. Two CIP-resistant Gram-negative bacilli were isolated from patients on CIP + COL compared with 13 NEO-resistant Gram-negative bacilli from patients on NEO + COL. Following a subsequent decade of unchanged use of this prophylactic strategy in neutropenic patients, a 2-year follow-up study between 1 January 1998 and 31 December 1999 showed 66 significant infections during 350–400 neutropenic episodes. Eight of the 111 (7·2%) isolates were with ciprofloxacin-resistant organisms, involving 2% of the neutropenic episodes, indicating that the strategy of combining colistin with ciprofloxacin has been effective in the prevention of Gram-negative sepsis in neutropenic patients without the emergence of significant resistance despite widespread concurrent hospital and community use of the quinolones.
Infection has been, and remains, the major cause of morbidity and mortality in neutropenic patients (Young, 1983; Rossi & Klastersky, 1996,Riley et al, 1999; Hargrave et al, 2001). As the outcome of therapy for acute leukaemia has improved, the number of patients experiencing prolonged (> 10 d), profound (neutrophils < 0·1 × 109/l) neutropenia has also increased (Schimpff, 1980; Gratwohl et al, 1998; Jagarlamundi et al, 2000). Patients with leukaemia who receive chemotherapy usually have concurrent gut and respiratory tract mucosal damage, and percutaneous vascular access catheters, all of which can be portals of entry for infecting organisms. The majority of bacterial infections are caused by endogenous organisms or acquired Gram-negative bacteria which colonize the gastrointestinal tract (Schimpff et al, 1972).
Over the past three decades, numerous studies have examined ways of suppressing potential pathogens already colonizing the patient and reducing the acquisition of new organisms. Attention has focused chiefly on the use of laminar air flow systems (O'Donnell et al, 1994) and attempts to achieve either total gut decontamination using oral antibiotics (Storring et al, 1977; Watson et al, 1982) or ‘selective’ gut decontamination based on Van de Waaij's concept of colonization resistance (van der Waaij & Berghuis, 1974). Some authors (Wade et al, 1981; Watson et al, 1982) have suggested that, despite colonization resistance, systemic prophylaxis by absorbable cotrimoxazole prevented infection more effectively than the simple suppression of intestinal bacteria by non-absorbable agents.
Ciprofloxacin (CIP), a 4-quinolone antibiotic, is active against most Gram-negative and some Gram-positive bacteria (Fass, 1983). It rapidly eliminates aerobic Gram-negative organisms from the gut after oral administration and is also absorbed, reaching systemic therapeutic levels within 1–2 h (Bergan et al, 1988). Ciprofloxacin has little effect on the anaerobic flora of the gut and may preserve colonization resistance (Rozenberg-Arska et al, 1985).
Several small studies have investigated ciprofloxacin in the prevention of bacterial infection in neutropenic patients, but the emergence of ciprofloxacin-resistant organisms has been a concern (Rozenberg-Arska et al, 1985; Dekker et al, 1987; Young, 1987; Cometta et al, 1994; Yeh et al, 1999). It has previously been suggested that it is possible to limit the emergence of resistant strains by the concomitant use of another antibiotic (Rozenberg-Arska et al, 1983). We previously conducted an in vitro study (Mannan et al, 1988) which excluded antagonism between ciprofloxacin and colistin (COL) when tested against strains of Pseudomonas aeruginosa isolated from neutropenic patients, suggesting a possible role in prophylaxis for this regimen. Hence, in the comparative trial, the combination of ciprofloxacin with colistin was evaluated by comparing it with what, at that time, was a standard prophylactic regimen of neomycin (NEO) plus colistin (Storring et al, 1977). Subsequent to this trial, the intention was to put the best arm of the study into practice and assess the risk of emergence of resistant organisms over the next decade of use in a unit managing approximately 350–400 neutropenic episodes in patients with haematological malignancy per annum.
- Top of page
- Patients and methods
Selective oral gut decontamination is now widely, but not universally, used for the prevention of infections caused by Gram-negative bacilli in neutropenic patients, although its role is controversial (Young, 1987), and most studies have only shown a significant benefit when combined with a protective environment. It is, perhaps, now more widely used in the intensive care setting, in which meta-analysis has shown a reduction in ventilator-associated pneumonia and in mortality, particularly in acute trauma patients (D'Amico et al, 1998).
A number of antimicrobial agents such as colistin, cotrimoxazole and the 4-quinolones have the capacity to suppress gut enterobacteriaceae without reducing the anaerobic flora (Rozenberg-Arska et al, 1985). Many studies have investigated the use of cotrimoxazole, but there has been concern about its potential for marrow suppression (Wade et al, 1983) and the incidence of adverse reactions (Watson et al, 1982). More recent studies show faster neutrophil recovery with ciprofloxacin prophylaxis compared with cotrimoxazole prophylaxis (Imrie et al, 1995). Further trials have demonstrated encouraging results for the quinolones with a reduction in the incidence of Gram-negative sepsis and febrile episodes (Dekker et al, 1987; Karp et al, 1987; Warren et al, 1990; GIMEMA Infection program, 1991; Schmeiser et al, 1993; Lew et al, 1995; Ford et al, 1998) and further confirmation of its safety and efficacy is provided by this study.
The CIP + COL regimen provided better prophylaxis against bacterial infection than NEO + COL, with a greater number of patients on CIP + COL remaining afebrile during their period of neutropenia. This study also provides convincing evidence of the reduction in duration of fever and use of parental antibiotics with CIP + COL prophylaxis. In part, this is related to the difference in the number of soft tissue and pulmonary infections between the two regimens. Patients on NEO + COL had more of these types of infection which took longer to respond and necessitated longer periods on parenteral antibiotics. CIP is well absorbed and achieves good tissue levels (Fass, 1983), suggesting that it should be a good agent for preventing infections originating in the respiratory tract and skin.
Only 14 patients on CIP + COL had bacteriologically documented infections compared with 39 patients on NEO + COL. It has been argued that the use of oral absorbable agents could prevent the recovery of organisms from culture specimens and give rise to misleading results. However, this does not appear likely as there was an equivalent number of non-bacteriologically documented pyrexial episodes between the two groups, evidence reinforced by the observation of fewer febrile episodes in the CIP + COL group.
The CIP + COL regimen was effective in preventing infections with Gram-negative bacilli and S. aureus. There were no S. aureus infections in the CIP + COL arm and only two patients developed infections with Gram-negative bacilli. Both were Hickman catheter exit site infections which responded following treatment with IV ceftazidime and amikacin and removal of the Hickman catheters. In contrast, 25 patients on NEO + COL developed bacteraemias, soft tissue or pulmonary infections with Gram-negative bacilli or S. aureus. These results reflect the antibacterial spectrum of the antibiotics and the local microbial ecology. In our institution, Pseudomonas aeruginosa had previously accounted for about 30% of Gram-negative infections, with 59% mortality, in neutropenic patients and there had also been a high incidence of S. aureus infection. The spectrum of infection in neutropenic patients is changing (Kibbler & Prentice, 1999) and factors influencing this include prophylactic antimicrobials, the use of long-term central venous catheters (Lowder et al, 1982) and mucosal damage following intensive cytotoxic therapy (McWhinney et al, 1991). It is important that the choice of strategies for antimicrobial prophylaxis in the neutropenic patient is guided by the prevalent pathogens and their antimicrobial susceptibility.
In our institution, 50–55% of E. coli isolated from our neutropenic patients were resistant to neomycin. This may account for the difference in rates of isolation of resistant Gram-negative bacilli between the two regimens. NEO-resistant Gram-negative bacilli were isolated from 13 patients on NEO + COL compared with only two CIP-resistant Gram-negative bacilli from patients on CIP + COL. However, only one of the eight gut isolates of NEO-resistant enterobacteriaceae caused bacteraemia.
As illustrated by the long-term surveillance study, the emergence of CIP-resistant Gram-negative bacilli during therapy remains an infrequent event in our patient population. Resistant strains have been reported from a number of centres (Cometta et al, 1994; Kern et al, 1994; Yeh et al, 1999) and have led to concerns that quinolone prophylaxis should be discontinued. Similar problems were encountered in the past with co-trimoxazole prophylaxis and it was found that the acquisition rate and number of infections due to resistant enterobacteriaceae may be reduced by the addition of a concomitant antibiotic such as colistin (Rozenberg-Arska et al, 1983). Although the dose of colistin used in the initial study was lower than the currently recommended daily dose of 4·5–9 MU, it may have contributed to the prevention of selection of resistant organisms. We believe that the subsequent increased colistin dose has maintained this effect.
In our initial study, neither of the two cases of Gram-negative infection in patients receiving CIP + COL were with ciprofloxacin-resistant organisms and two out of five colonizing organisms were resistant. This compares favourably with a recent published meta-analysis of trials mostly performed during that era, showing an incidence of infections with quinolone-resistant Gram-negative organisms of 3·0% (Engels et al, 1998). After 10 years further use of our regimen, the incidence of infections with ciprofloxacin-resistant Gram-negative organisms is 1·2% and there has been no increase in the incidence of colonization with ciprofloxacin-resistant Gram-negative organisms (31% versus 40% of isolated Gram negative).
Both regimens were well tolerated. Fewer patients on CIP + COL had diarrhoea, which may relate to the lower use of IV antibiotics for patients in this arm. In general, compliance was good with both regimens, a factor of great importance with regard to the usefulness of prophylactic therapy. Another useful effect of antibiotic prophylaxis is in the reduction of acute graft-versus-host disease (GvHD) following allogeneic bone marrow transplantation (allo-BMT) (Beelen et al, 1999). The hypothesis is that the intestinal bacterial microflora are involved in the pathogenesis of GvHD through the LPS (lipopolysaccharide)-mediated enhancement of the cytokine storm (Ferrera, 1993) and that the addition of anti-anaerobic agents such as metronidazole to ciprofloxacin might reduce the incidence of GvHD. The randomized trial carried out by Beelen et al (1999) demonstrated a reduced incidence of severe acute GvHD but no effect on chronic GvHD or survival.
In conclusion, ciprofloxacin provides effective prophylaxis against Gram-negative infection in adults. Our study shows that careful use, combining it with colistin in the setting of good infection control practices to prevent the spread of imported resistance, has maintained its effectiveness. Thus, in the absence of any superior strategy, this approach continues to be used and recommended.