Oral Antimicrobial Prophylaxis in Bone Marrow Transplant Recipients: Randomized Trial of Ciprofloxacin versus Ciprofloxacin-Vancomycin

Abstract

The optimal oral antimicrobial prophylactic regimen for bone marrow transplant recipients remains to be elucidated. We randomized 84 patients to receive either oral ciprofloxacin or ciprofloxacin plus vancomycin at hospital admission. Patients were monitored for bacteremias and clinical parameters, and stool and throat swab surveillance cultures were performed. The addition of vancomycin resulted in a significant decrease in the frequency of patients with surveillance cultures positive for coagulase-negative staphylococci (stool cultures, 44 versus 23%; throat swab cultures, 37 versus 19%) and alpha-hemolytic streptococci (throat swab cultures, 90 versus 60%). The frequencies of positivity for Candida spp. and gram-negative organisms on surveillance cultures were comparable. Despite these results, no differences in the incidences of bacteremias (12 of 41 versus 12 of 43 patients) or clinical parameters such as number of days to first fever, total number of febrile days, length of stay, and number of transfusions could be demonstrated. Because of a lack of efficacy of vancomycin and emerging problems with vancomycin-resistant isolates, vancomycin should not be used in oral antimicrobial prophylaxis regimens.

Granulocytopenic fever and bacteremia are important complications in bone marrow transplant recipients. Several approaches that include prophylactic antibiotic regimens are often used to minimize infection (8, 15). Such antibiotic regimens must reach the sources of potential bacterial invasion such as the gut and oral mucosa and would ideally be oral regimens to reduce cost and inconvenience, especially if the patients are outpatients. In recent years older regimens with more toxic multiple oral nonabsorbable antibiotics have largely been replaced by fluoroquinolones (12, 14, 18). The newer regimens, however, have been associated with the emergence of bacteremias caused by gram-positive organisms, with potentially serious morbidity and mortality (3, 6).

In order to address the problem of the emergence of gram-positive organisms in transplant recipients, randomized studies of quinolones with or without oral erythromycin (21), roxithromycin (11), and penicillin (10) have been reported. While the added drug was usually effective in decreasing the numbers of cases of streptococcal bacteremia, bacteremia caused by other gram-positive organisms, especially coagulase-negative staphylococci (CoNS), and clinical parameters were not consistently affected. Archimbaud et al. (1) conducted a randomized study comparing pefloxacin-vancomycin with gentamicin-colistin-vancomycin. Patients receiving the quinolone plus vancomycin had significantly fewer infections caused by gram-negative and gram-positive organisms, including fewer line infections and fewer infections caused by isolates of CoNS. These results suggest that the combination of a quinolone and vancomycin represents an acceptable and possibly optimal oral prophylactic regimen for the prevention of bacteremias in transplant recipients.

In order to clarify the value of oral vancomycin when it is added to a quinolone, we began a randomized trial of oral antibiotic prophylaxis in bone marrow and peripheral stem cell recipients receiving ciprofloxacin. During the course of the study, it became more apparent that the administration of oral prophylactic antibiotics may be associated with the emergence of resistant organisms, including vancomycin-resistant enterococci (16, 20). Since many institutions had not established guidelines for vancomycin use, however, it was felt that continuing the trial was justified in order to establish whether oral vancomycin had any clinical benefit when used in regimens for selective decontamination of the digestive tract.

MATERIALS AND METHODS

Patients.

LDS Hospital is a 520-bed private, acute-care referral hospital located in Salt Lake City, Utah. The patient population consisted of 84 patients admitted for allogeneic or autologous transplantation between November 1991 and July 1996. During the study period, we saw 10 additional patients who were potentially eligible but who were not entered into the study because of patient refusal or because of our failure to randomize the patients and obtain a signed consent prior to the administration of the first dose of oral antibiotic. No patients entered into the study were excluded from the analysis. Patients were stratified for type of transplant (allogeneic or autologous) and treatment with total body irradiation. Patients with fluoroquinolone allergy were not eligible. Tunneled central lines were placed in all patients either at admission or prior to admission for the purpose of collecting mobilized peripheral blood stem cells. The treatment protocol was approved by the institutional review board, and all patients signed informed consents.

Prophylactic antibiotics.

Consenting patients were assigned to one of the treatment groups by a randomized block design. They received either ciprofloxacin at 750 mg orally (p.o.) twice a day with vancomycin at 250 mg p.o. three times a day or ciprofloxacin alone. Vancomycin was administered as a liquid, with the patients being encouraged to swish the liquid in the mouth and gargle before swallowing. Oral antibiotics were begun on the day of admission and were continued until either systemic broad-spectrum antibiotics were started or the patient was afebrile with a rising absolute granulocyte count exceeding 500/μl. Systemic antibiotics were initiated at the discretion of the attending physician, usually on the first day of granulocytopenic fever, and most often consisted of monotherapy with imipenem-cilastatin. Patients also routinely received fluconazole and acyclovir daily p.o. (or intravenously when they were unable to take the drugs p.o.) beginning on the day of admission.

Laboratory studies.

Blood cultures were performed for first fever, and blood samples for culture were usually obtained at least daily in the event of continuing fever. Most often, two sets of blood for culture were drawn through the central line. Positive cultures were defined according to guidelines from the Centers of Disease Control and Prevention, namely, an appropriate clinical picture and the retrieval of more than one isolate of the common skin flora (i.e., isolates differing in time or location) (7). Thus, single cultures with CoNS (six patients) were considered contaminants. Surveillance samples for culture (stool and throat swab specimens) were obtained on admission and weekly until discharge from the hospital.

Statistical methods.

The primary endpoint of this study was the development of bacteremia caused by a gram-positive organism. The study defined a significant reduction in the incidence of bacteremia caused by gram-positive organisms as a result of the addition of vancomycin as a two-thirds decrease. Since our previous experience (6) had shown that five of nine patients receiving norfloxacin developed bacteremia caused by gram-positive organisms, we estimated the control incidence to be 40%. The study size was designed to detect the desired reduction at an 80% power and a two-tailed statistical significance of 0.05. Secondary endpoints included the number of days to first granulocytopenic fever (>38°C), total number of febrile days, number of days in the hospital, changes in the frequency of gram-positive isolates in surveillance cultures, and the incidence of diarrhea associated with Clostridium difficile toxin. Data between the two arms of the study were compared by a two-tailed independent samples t test or Pearson’s chi-square test.

RESULTS

The characteristics of the 84 patients who were randomly assigned to the two study groups are summarized in Table 1. There were no significant differences between the study arms in type of transplant (allogeneic versus autologous), stem cell source (bone marrow versus blood), the administration of total body irradiation, or primary disease. Nonstratified factors which might have affected some of the study endpoints such as the number of days of absolute granulocytopenia, the number of transfusions, and the length of hospital stay were also well balanced (Table 2).

TABLE 1.

Patient characteristics

Characteristic	Ciprofloxacin group	Ciprofloxacinvancomycin group
No. of patients	41	43
Stem cell source (no. of patients)
Autologous marrow	4	5
Allogeneic marrow	9	10
Autologous peripheral stem cell	22	21
Allogeneic peripheral stem cell	6	7
Total body irradiation (no. of patients)	15	18
Primary tumor at the indicated site or disease (no. of patients)
Breast	17	14
Ovary	2	3
Sarcoma		2
Non-Hodgkin’s lymphoma	6	10
Hodgkin’s disease	1	1
Acute myelogenous leukemia	5	7
Acute lymphocytic leukemia	1	2
Chronic myelogenous leukemia	3	4
Chronic lymphocytic leukemia	2
Myeloma	2
Aplastic anemia	1
Myelodysplasia	1

TABLE 2.

Clinical parameters for the two study groups^a

Parameter	Ciprofloxacin group	Ciprofloxacinvancomycin group	P
Total no. of days with AGC of <100	9.3 (1.2)	11.2 (1.6)	0.34
Total no. of days with AGC of <500	13.9 (1.6)	14.9 (1.8)	0.69
Total no. of days with AGC of <1,000	16.5 (2.1)	16.6 (1.9)	0.99
No. of RBC transfusions (units)	9.5 (1.6)	8.8 (1.6)	0.76
No. of platelet transfusions	13.3 (3.4)	10.9 (2.5)	0.58
Length of stay (days)	30.9 (2.9)	28.7 (2.2)	0.56
Total no. of febrile days	9.5 (1.5)	11.9 (1.5)	0.26
No. of days to first fever	7.4 (1.2)	6.2 (0.86)	0.38

^aAGC, absolute granulocyte count; RBC, erythrocyte. Values are means (standard errors). 

Eight study patients died during the first 100 days following transplantation. These included three in the ciprofloxacin group (poor engraftment, graft-versus-host disease, and thrombotic thrombocytopenic purpura) and five in the ciprofloxacin-vancomycin group (rejection, hemorrhage, veno-occlusive disease of the liver, recurrent leukemia, and interstitial pneumonia). Among the study patients there was no mortality due primarily to bacterial infection.

Table 3 summarizes the results of the blood cultures for the study patients. Both regimens were effective in preventing bacteremia; i.e., <10% of blood cultures were positive. The most common organisms were CoNS. There was no difference between the groups in the rates of infections caused by gram-negative or gram-positive bacteria or fungi. No vancomycin-resistant isolates were recovered from either group.

TABLE 3.

Blood culture results

Characteristic	Ciprofloxacin group	Ciprofloxacinvancomycin group	P
No. of patients	41	43
Transplant day with first positive culture result	10.6 (3.1)a	9.1 (3.4)	0.74
No. of patients with positive culture results	12	12	NSb
No. of patients with bacteremia caused by the following:
Gram-positive bacteria	10	7	0.35
Alpha streptococci	2	3
CoNS	5	3
Staphylococcus aureus	1
Micrococcus spp.	1
Diphtheroid spp.	1
Bacillus spp.		1
Gram-negative bacteria	2	2	NS
Enterobacter cloacae	1
Klebsiella pneumoniae	1
Pseudomonas aeruginosa		1
Fusobacterium necrophorum		1
Unknown		1
Candida krusei	2	2	NS
No. of patients C. difficile toxin positive	2	2	NS

^aValues are means (standard errors). 

^bNS, not significant. 

The incidence of diarrhea associated with C. difficile toxin was low in each group, possibly because of the activity of ciprofloxacin (12). However, at our institution C. difficile-associated diarrhea is relatively uncommon in both transplant and nontransplant patients. The addition of vancomycin did not diminish the low incidence seen in those receiving ciprofloxacin alone. The addition of vancomycin also did not affect meaningful clinical parameters. The total number of days of fever and the number of days to the onset of the first granulocytopenic fever were identical in the two groups. Similarly, the requirement for blood products and length of hospital stay did not significantly differ between the two groups (Table 2).

Table 4 summarizes the results of the surveillance cultures. There was a significant decrease in the number of patients with any stool culture showing CoNS in favor of the vancomycin group. A similar trend was seen for throat swab cultures. The incidence of alpha-hemolytic streptococci was lower for the throat swab cultures of patients receiving vancomycin. No differences in the frequencies of Candida spp. were seen.

TABLE 4.

Surveillance culture results

Organism	No. of patients with a positive culture resulta
	Stool sample			Throat swab
	C (n = 41)b	CV (n = 43)	P	C (n = 41)	CV (n = 43)	P
CoNS	18	10	0.045	15	8	0.06
Alpha streptococci	10	9	0.71	37	26	0.002
Candida spp.	22	26	0.53	19	14	0.20
Gram-negative bacteria	NA	NA		6	4	NS

^aC, ciprofloxacin group; CV, ciprofloxacin-vancomycin group; NA, not available; NS, not significant. 

^bn indicates total number of patients. 

DISCUSSION

The emergence of infections caused by gram-positive organisms, especially bacteremia due to streptococci and CoNS, in bone marrow transplant recipients receiving oral prophylaxis with quinolones is well known. We recognized this problem at our own institution when we changed from an oral prophylactic regimen containing vancomycin to one containing norfloxacin (6). Subsequently, we noted an increase in norfloxacin-resistant non-beta-hemolytic streptococcal bacteremias that sometimes were associated with systemic inflammatory response syndrome and, occasionally, mortality. Therefore, we sought to evaluate prophylactic regimens with improved coverage against gram-positive organisms.

Our results with oral vancomycin generally mirror the outcomes of other studies that have evaluated oral coverage against gram-positive organisms, in addition to quinolones, in bone marrow transplant recipients. The European Organization for Research and Treatment of Cancer randomized patients to receive penicillin V or placebo, in addition to pefloxacin. The addition of penicillin resulted in a significant decrease in the total number of cases of bacteremia due almost entirely to decreases in streptococcal infections, but the incidences of bacteremia caused by CoNS and other bacteria were not affected. Intravenous penicillin has been reported to produce a similar effect on the incidence of bacteremia (5). Whether the addition of macrolide antimicrobial agents is beneficial has also been disputed (11, 21).

The patients in the present study developed a lower incidence of bacteremia caused by any organism, including streptococcal bacteremia, than patients from our own as well as some other centers. In our previous study (6) of patients receiving norfloxacin orally, we found that all streptococcal isolates in the bloodstream were resistant to norfloxacin but sensitive to ciprofloxacin, suggesting that norfloxacin is inferior for prophylaxis of streptococcal infection. Excess streptococcal infections were also observed in granulocytopenic patients receiving norfloxacin in the studies of Bow et al. (4) and Broun et al. (5). However, a potential confounding factor is the relatively frequent use of antibiotics before admission for transplantation such as during granulocytopenia from previous therapy and stem cell mobilization and for line insertion prophylaxis.

In addition, bacteremias due to CoNS have varied in frequency among institutions and have been difficult to control by oral prophylactic regimens. In our study, the rate of recovery of CoNS from cultures of stool and throat swab specimens was decreased by oral vancomycin. This may have resulted in a decrease in the number of cases of bacteremia caused by CoNS, although the numbers of patients in the two study groups are too small for definite conclusions. Nevertheless, the decrease in the numbers of gram-positive organisms isolated from the oropharynges of patients swishing and gargling with vancomycin is of interest since this result suggests efficacy with topical administration. Bacteremias caused by CoNS remain a significant and unresolved clinical problem in which sources other than the gut, especially central lines, play an important role.

While oral prophylactic antibiotics have decreased the incidence of bacteremias, they have also been associated with the development of resistant bacteria such as fluoroquinolone-resistant gram-negative organisms (22) and vancomycin-resistant enterococci (16, 20). Furthermore, minimization of antibiotic use may improve resistance patterns (19). Since the addition of prophylaxis for infections caused by gram-positive organisms to oral quinolone regimens has produced no consistent improvement in clinical outcomes, we have discontinued its routine use and added coverage against gram-positive organisms at the onset of fever, clinical deterioration, or a positive blood culture.

Our study suggests that the use of oral vancomycin in prophylactic regimens can no longer be justified both because of its failure to improve clinical outcomes and because of concerns about the emergence of resistance. Thus, our results lend further support to recent recommendations that vancomycin not be used for selective decontamination of the gastrointestinal tract (9). In addition to drug resistance, intravenous vancomycin may be associated with additional toxicities such as an increase in the incidence of veno-occlusive disease of the liver and other adverse drug events (2, 13, 17).