- Email: [email protected]
A prospective, randomized comparison of ceftazidime and ciprofloxacin as initial empiric therapy in neutropenic patients with fever
A Prospective, Randomized Comparison Ceftazidime and Ciprofloxacin as Initial Empiric Therapy in Neutropenic Patients with Fever KATHERINE
F. BAYSTON, F.R.C.P.,’ SUEWANT, Ph.D,
JONATHAN COHEN,
This study developed further clinical experience in using a single agent (“monotherapy”) as empirical treatment for neutropenic patients with fever, and compared the safety and toxicity of two candidate agents, ceftazidime and ciprofloxatin. A prospective, randomized, single-center efficacy and safety comparison was conducted of intravenous ciprofloxacin, 200 mg every 12 hours, and ceftazidime, 2 g every eight hours, as initial empirical therapy in neutropenic patients with fever. Regimens were modified as necessary, guided by laboratory results and/or the clinical condition. Response was evaluated at 72 hours and at the end of the neutropenia. Toxicity was evaluated by regular clinical examination and laboratory investigations. A total of 43 patients with 51 febrile neutropenic episodes were enrolled into the study and randomly assigned to one of the two regimens. Five episodes were excluded from evaluation of efficacy because of protocol violations, leaving 46 evaluable episodes (21 ciprofloxacin, 25 ceftazidime). The two groups were well matched for risk factors for infection. There were no differences between the two groups in response rates either at 72 hours or at the end of neutropenia, although in the vast majority of patients some modification of the initial therapy was required. No patients died of uncontrolled bacterial infection. Superinfection with gram-positive cocci (often streptococci) was seen primarily in bone marrow transplant recipients who had been randomly assigned to receive ciprofloxacin. This study demonstrated that, in certain circumstances, a single antibiotic can be used successfully as initial empirical therapy in febrile neutropenic patients. In this study, ceftazidime and ciprofloxacin were generally of equal efficacy, but there appeared to be an increased incidence of streptococcal superinfection in bone marrow transplant recipients who received ciprofloxacin.
From the Infectious Diseases Unit. Departments of Bacteriology and Medicine, Hammersmith Hospital and Royal Postgraduate MedIcal School. London, United Kingdom. This work was supported by a grant from Bayer (UnIted Kingdom). Requests for reprints should be addressed to Dr. Jonathan Cohen. Hammersmith Hospital, Du Cane Road, London W12 ONN, Unlted Kingdom ‘Katherine F. Bavston is an Overseas Research Fellow of the MedIcal Research Council of New Zealand. tJonathan Cohen IS a Weilcome Trust Senior Lecturer.
of
MSc., F.R.C.P.t London, Umted Kfngdom
M
any clinical studies conducted over- the last 15 years have clearly established the benefit of the early empirical use of antibiotics in febrile neutropenic patients [1,21. The success of this approach is attributable, in part, to a knowledge of the epidemiology of infection in these patients, and, in part, to the growing number of potent, bactericidal antibiotics that have become available. Recent changes in both of these factors have led to a re-evaluation of the choice of antibiotics to be used for empirical therapy, notably, the suggestion that a single agent (so-called monotherapy) might be as effective as the traditional combination of a penicillin coupled with an aminoglycoside [3]. Ceftazidime is an extended spectrum cephalosporin that has been used successfully as a single agent for empirical therapy [4-61. Another. drug that merits consideration in this role is ciprofloxacin, a fluoroquinolone antibiotic with a favorable in v&o profile and the potential advantage of both oral and parenteral formulations. We have, therefore, carried out a prospective, randomized comparison of these two drugs as initial empirical therapy in a group of febrile neutropenic patients.
METHODS Entry Criteria This single-center trial was conducted at the Hammersmith Hospital, London, United Kingdom. All enrolled patients were receiving inpatie.nt treatment and met the following study criteria: age at least 18 years; presence of fever (defined as a temperature of 38°C on two consecutive occasions two hours apart or a single recording of 39°C or higher); and neutropenia (defined as a total granulocyte count of 1 x log/liter or less). Exclusi’on criteria included known allergy to a trial drug; pregnancy or lactation; serum creatinine level greater than 3 mgidl; suspected central nervous system infection; infection caused by an organism known to be resistant to either of the trial drugs; and, finally, the use of any effective antimicrobial therapy active against the infecting pathogen within 72 hours of trial entry, other than antibiotics used for selective alimentary tract contamination. No patient could enter the study more than once during the same episode of neutropenia.
TRIAL DESIGN AND TREATMENT DEFINITION Patient Allocation Patients were randomly allocated to one of the two treatment groups, using a computer-generated randomization. Treatment was not blinded.
November 30, 1989
The American Journal of Medicine
Volume 87 (suppl 5Aj
5A-269s
SYMPOSIUM
ON CIPROFLOXACIN
/ BAYSTON ET AL
Drug Treatment Patients with normal renal function received either ciprofloxacin 200 mg every 12 hours by intravenous infusion, or ceftazidime 2 g every eight hours as an intravenous bolus. Dosages for patients with impaired renal function were modified appropriately. Drug levels were not routinely monitored. All patients received gut decontamination from the time of admission to hospital. This generally consisted of co-trimoxazole 960 mg twice daily and oral nonabsorbable antifungal agents. Colistin 1.5 mU every six hours was added if resistant organisms were found in surveillance culture specimens. In addition, bone marrow transplantation patients received prophylactic acyclovir. Modification of therapy, defined as the use of any other antimicrobial agent, including the use of granulocyte transfusions, either in addition to or in substitution for the assigned drug regimen, could be made after study enrollment based on either clinical deterioration or documentation of antibiotic resistance. Duration of Therapy Study patients were first assessed at 72 hours after being randomly assigned to therapy. Patients who were judged to have had responses to treatment continued therapy with that agent until neutropenia had resolved or until they had been afebrile for at least four consecutive days and there was evidence of satisfactory control of focal infection. For those patients who had not had responses to treatment? therapy was altered as necessary and treatment with the study drug continued or discontinued at the discretion of the primary physician. Study therapy was stopped if an allergic reaction developed. All study patients were observed until resolution of infection and neutropenia, or death. Clinical and Laboratory Monitoring Prior to starting treatment, a complete history was obtained and a complete physical examination was performed on all patients. Patients were examined daily and the appearance of rashes or other possible allergic manifestations were noted. All patients had twice-weekly chest radiographic examinations from the time of admission, regardless of symptoms. Blood counts were performed daily and biochemistry profiles obtained at least on alternate days during neutropenia. Additional diagnostic tests were performed as required. Bacteriologic Evaluation Twice-weekly surveillance culture specimens of feces, urine, and throat gargles were obtained from all patients. Prior to commencing trial therapy, blood culture specimens and culture specimens of any suspected sites of infection were obtained. In patients with an indwelling venous catheter (most commonly a Hickman-Broviac) at least one sample was obtained through the catheter and, where possible, another from a peripheral vein. Blood culture specimens were obtained and paired samples were cultured aerobically and anaerobically using the Bactec system. Follow-up blood culture specimens were taken daily during persistent fever. Documented sites of infection were recultured whenever possible within 72 hours of clinical response and/or discontinuation of antibiotic therapy. Bacterial isolates were identified by standard bacte5A-270s
November 30, 1989
The American Journal of Medicine
riologic techniques. Antibiotic susceptibility was determined in the first instance by disk diffusion and subsequently by the microtiter technique for minimal inhibitory concentrations.
CLASSIFICATION OF INFECTION Febrile episodes were assigned to one of four groups. Episodes accompanied by definite clinical signs and symptoms of infection plus microbiologic confirmation were termed microbiologically documented infections, with or without bacteremia, based on blood culture specimen results. Urinary tract infections were diagnosed when a midstream urine specimen contained more than lo5 colony-forming units/ml. Episodes accompanied by definite clinical features of infection, but without microbiologic proof, were termed clinically documented infections. Pneumonia was diagnosed by the presence of a new pulmonary infiltrat,e not attributable to a noninfectious cause. All other febrile episodes were classified as either possible or doubtful infections. Superinfection was taken to be the isolation of an organism different from that initially causing fever (or the first isolation of an organism after a period of response) during the course of therapy. Evaluation of Response Response was evaluated at 72 hours and again at the end of therapy. At the first evaluation, outcome was classified in one of three categories: (1) success, defined as a return of temperature to normal (i.e., temperature less than 37.5”C), and maintenance of an afebrile state for at least 48 hours thereafter; (2) improvement, defined as temperature less than 38°C and clinically improved without modification of trial therapy; or (3) failure! defined as death from infection or clinical deterioration necessitating alteration of therapy. At the second evaluation, episodes were classified as: (1) success, defined as an episode in which recovery from infection and neutropenia was achieved without modification of therapy; (2) success with modification, defined as the patient recovering from an episode of infection but requiring modification of the assigned regimen; and (3J failure, defined as death occurring during the febrile neutropenic episode. Toxicity Renal or hepatic abnormalities were considered to be related to the study drugs in the absence of other attributable causes. Patients in whom features of septic shock developed, who received cyclosporin A or amphotericin B, or in whom hepatic veno-occlusive disease or graft-versus-host disease involving the liver developed during the period of therapy were excluded from assessment of toxicity. Statistics Fisher’s exact test was used to compare the distribution of febrile episodes and the outcome of therapy, and the Student t test was used to compare duration of neutropenia in the two study groups.
RESULTS Characteristics of the Study Population Over a 24-month period, 43 patients with 51 neutropenic febrile episodes were randomly assigned to receive single-agent empirical therapy with either ciprofloxacin or ceftazidime. Five episodes were excluded
Volume 87 (suppl 5A)
SYMPOSIUM
from evaluation because of trial violations, leaving 46 evaluable episodes in 40 patients. Of these, 21 were treated with ciprofloxacin and 25 with ceftazidime. As shown in Table I, the patients in each group were comparable with respect to age, sex, duration of neutropenia, and underlying diagnosis. Neutropenic episodes generally occurred in the context of bone marrow transplantation for chronic myeloid leukemia or acute myeloid leukemia (57 percent overall), or chemotherapy for acute leukemia (41 percent). An additional patient in the ciprofloxacin group became neutropenic after therapy for blast transformation of chronic myeloid leukemia. Classification of Episodes There were no statistically sigtiificant differences in the distribution of febrile episodes between the two groups. Microbiologically documented episodes occurred in seven of 21 episodes (33 percent) in the ciprofloxacin group, all of which were baeteremias, and in 16 of 25 (64 percent) treated episodes in the ceftazidime group (p = 0.07). Of the latter, 11 were solely bacteremic episodes, four were urinaky tract infections alone, and, in one patient, bacteremia and urinary tract infection (caused by different organisms) occurred together. The remaining episodes in both groups were classified as clinically documented episodes. Outcome Evaluatian There was no difference in time to defervescence between the two groups (ciprofloxacin, 5.05 + 4.18 days; ceftazidime, 4.6 i 3.34 days; p = 0.7). At 72 hours, the outcome for episodes treated with ciprofloxacin was no different from that for episodes treated with ceftazidime. Success occurred in seven of 21 (33 percent) versus 10 of 25 (40 percent), improvement in three of 21 (14 percent) versus three of 25 (12 percent), and failure in 11 of 21 (52 percent) versus 12 of 25 (48 percent) recipients of ciprofloxacin versus ceftazidime, respectively. No patients died during this period. At the end of therapy, outcome was classified as success with modification in the majority of instances: 15 6f 21 (71 percent) episodes treated with ciprofloxatin and 16 of 25 (64 percent) episodes treated with ceftazidime (not significant). Success without modification was documented in only three of 21 (14 percent) ciprofloxacin patients and seven of 25 (28 percent) ceftazidime patients, and failure was documented in three of 21 (14 percent) and two of 25 (8 percent) patients, respectively. Details of bacterial isolates for each treatment group are given in Table II and sensitivity data (minimal inhibitpry concentrations) in Table III. There was no difference in outcome for microbiologically documented infections between the two treatment groups, either at 72 hours (p = 0.3) or at the end of therapy (p = 0.21). Bacterial .superinfections (six patients) occurred predominantly in bone marrow transplant recipients (five of si?i patients) and most commonly in recipients of ciprofloxacin: four streptococcal bacteremiai and one Capnocytophaga ochracea bacteremia. In the ceftazidime group, there was one episode of ClostAdimn d@Ele-associated diarrhea. Fungal superinfection occurred in one patient in each treatment group; both were bone marrow transplant recipients. November
’ TABLE
ON CIPROFLOXACINiBAYSTON
ET AL
I
Characteristics
of the Study
Population* Treatment Group
Characteristic
Ciprofloxacin
Number Sex (M:F] Ag;~;;rsl
Ceftazidime
7:14
Range Neutropenia Mean days from trial entry SD Dlagnosls BMP Acute leukemia Other
I&
256
264
18.86 13.39
17.88 9.24
12 8 1
if 0
BMT = bone marrow transplant: SD = standard deviation. *Evaluable patients only.
TABLE
II
Organisms
Isolated
in Primary
Infections* Treatment Group Ciprofloxacin
Ceftazidime
Single organisms Staphylococcus epidermldis
Streptococcus sanguis Streptococcus mitis Colynebacterium sp. C. ochracea Eschenchia cali E.co/d Pseudomonas aeruginosa Mlxed infections
S epiderm~d,siAcnetobactersa. S. epiderm/disiS. sangua/Streptococcus equrnus Corynebacterium sp.lE co/it
0
*Organisms are bacteremic isolates unless otherwe kkinary tract isolates.
TABLE
0 1 1
h stated
Ill
Sensitivities Infections
for All Bacterial and Superinfections
Isolates,
Including
Primary
MIC (fig/ml)* Organism S. epidermidis S.sanguis S.mitis Streptococcus equinus Streptococcus pneumonfae Corvnebacterium SD.
/
Number i 3 i 1
Ciprofloxacin
Ceftazidime
0.15 (0.06-0.25)t
23 [3.0-501 1.2 (0.05-3.0) 0.1 0.1
1.7 ‘;:y”] 0.01
4.0 ND
E CO//
8
P aeruginosa Acinetobacter sp. C ochracea
i
0.0;
2
8.0
0.01 (<.y.Ol)
iii (0.05$01)
i:;
MIC = minimal lnhlbitory concentration; ND = no data. *MinImal inhibitory concentrations are expressed as means. tNumbers In parentheses, ranges.
Skin rash possibly attributable to administration of a study antibiotic was seen in two patients from each group. Two patients, one from each study group, developed pneumonitis in addition to a skin rash. These events occurred in bone marrow transplant patients 30, 1989
The American Journal of Medicine
Volume 87 (suppl 5A)
5A-271s
SYMPOSIUM
ON CIPROFLOXACIN
/ BAYSTON ET AL
,ipe~g;,iP$”Ceftazldime +-----4
Vancomycln Amphoterun 1-1 Ciprofloxacin 1-1
+364,,
0
1
2
3
4
/
5
/I
6 7 Days
‘,
8
H
9
1,
IO
11
12
1
13
Figure 1. Temperature chart of a bone marrow transplant patient randomly assigned to receive ciprofloxacin. Antibiotics used are Indicated at the top of the chart. Bars, duration of antibiotic administration.
and were probably unrelated to antibiotic therapy. No hematologic or biochemical toxicity could be directly attributable to either of the study drugs. All but one death during the period of neutropenia occurred in bone marrow transplant patients. None of these patients had documented bacterial infection at the time of death. In the ciprofloxacin-treated group, two patients died with idiopathic pneumonitis and one with Aspergillus pneumonia. Of patients treated with ceftazidime, one died with an idiopathic pneumonitis and the other patient developed influenza B pneumonia following graft rejection and died of multi-organ failure.
COMMENTS These results provide further support for the view that a single antibiotic can be used safely as initial empirical therapy in febrile neutropenic patients. Five of the 46 evaluable patients died during neutropenic episodes, but in none was this due to documented bacterial infection. Thus, taking death due to bacterial infection as the ultimate outcome, ceftazidime and ciproflox&in were equally effective. Furthermore, no differences were apparent when comparing the frequency with which it was necessary to add to, or modify, the initial regimen. Seventy-two hours after starting therapy, approximately half the patients in each group (ciprofloxacin, 47 percent; ceftazidime, 52 percent) had responses such that no change of therapy was required. Taken with the remaining patients who did require alteration of their treatment, overall survival at the end of neutropenia was 18 of 21 (86 percent) patients in the ciprofloxacin-treated group and 23 of 25 (92 percent) patients’ in the ceftazidimetreated group. These results are broadly comparable with previous experience with ceftazidime monotherapy, which documented an initial response rate of 50 to 60 percent and an overall success rate of more than 95 percent [4,51. Bowever, a more detailed analysis of the results revealed one important difference between the two drugs in relation to the frequency of superinfections. These occurred almost exclusively in the subset of bone marrow transplant recipients (seven of 26 versus one of 20 patients). Four of the six episodes due to bacteria were caused by streptococci, and all four were in the ciprofloxacin group. The growing significance of gram-positive bacteremia as a complication of neutropenia has been exten-
5A-2725
November 30, 1989
The American Journal of Medune
sively documented [7,8]. The single most common group of isolates have been coagulase-negative staphylococci, but we and others have also drawn attention to the importance of Streptococcus viridans [9]. These bacteremias most likely originate in the oral cavity, and one may speculate that their predominance in the bone marrow transplant recipients reflects the damage to the oral mucosa caused by the more intense conditioning regimens used in these patients. It is of interest that streptococcal superinfection was a particular problem in the patients receiving ciprofloxacin, reflecting that drug’s modest i,n vitro activity against streptococci [lo]. Indeed, previous studies with ciprofloxacin in neutropenic patients have provided information on the development of resistance to ciprofloxatin among gram-positive cocci [ll!, prompting its use in combination with benzylpenicilhn [12] or vancomytin [131 to provide additional therapy. Our data confirm that it is frequently necessary to add anti-streptococcal or anti-staphylococcal antibiotics, but suggest that this can be guided by the clinical circumstances; not all patients need vancomycin. In addition, these data illustrate well the importance of careful definition of terms. If one takes the view of the European Organization for Research on Treatment of Cancer (EORTC) that the purpose of empirical therapy is to achieve “. . . a lasting return of temperature to the normal level . . . and resolution of all signs and symptoms, without addition of other antibiotics” [14], then an initial response rate of 30 to 40 percent at the first evaluation is clearly unacceptable. An alternative view [5] is that empirical therapy is designed only to prevent death in the early stages before a more specific, “tailored” antibiotic regimen is chosen based on bacteriologic results and the patients’ initial response. On this basis, the only cause of failure would be death due to infection during the period of neutropenia, yielding an overall success rate for this study of approximately 90 percent. Despite the apparently compelling logic of this reasoning, it can lead to problems of interpretation. Consider the clinical course of the patient illustrated in Figure 1, who was randomly assigned to receive ciprofloxacin but in whom multiple modifications of therapy were later required before a satisfactory clinical response was obtained. Using the criteria developed by Pizzo et al [51, we would be obliged to classify this patient as “success with modification,” yet the implication that the empirical use of ciprofloxacin had contributed to the successful outcome is difficult to justify and potentially misleading. On the other hand, the more rigorous criteria employed by the EORTC provides a clearer basis for comparison between different agents, but could hinder the assessment of novel approaches that do not lend themselves to this type of evaluation. These problems of definition should not be allowed to deflect attention away from the more important issue of the potential value of monotherapy. The results presented here add weight to the growing body of evidence that suggests that in selected circumstances a single drug can be used as empirical therapy for fever in neutropenic patients but underlines the importance of distinguishing carefully between risk groups within the general category of neutropenia. Both ciprofloxacin and ceftazidime are potential candidates, and merit further study.
Volume 87 (suppl 5A)
SYMPOSIUM
ACKNOWLEDGMENT We are grateful to the staff of the Haematology Department, Hammersmith Hospital, for allowing us to study patients under their care. REFERENCES 1. Schlmpff SC, Satterlee W, Young VM, Serprck A: Emprric therapy with carbenrcrll~n and gentamrcrn for febrile patrents wrth cancer and granulocytopenra. N Engl J Med 1971; 284: 1061-1065. 2. Schimpff SC: Emprric anhbrotic therapy for granulocytopenrc cancer patrents. Am J Med 1986; 80 (suppl 5C): 13-20. 3. Hathorn JW, PIZZO PA: Is there a role for monotherapy wrth beta-lactam antibiotrcs rn the inihal emoirical manaeement of febrile neutrooenic cancer oatrents? J Antrmrcrob Chemother 1986; 17 (sup;1 A): 41-54. 4. Donnelly JP, Marcus RE, Goldman JM, et a/: Ceftazldime as first-lrne therapy for fever in acute leukaemra. J Infect 1985; 11. 205-215. 5. Pizza PA, Hathorn JW, Hiemenz J, era/: A randomrzed trial comparrng ceftazidime alone wrth combrnation antibiotic therapy in cancer patients with fever and neutropenia. N Engl J Med 1986; 315, 552-558. 6. De Pauw B, Wrlliams K, De Neeff J. eta/: A randomrzed prospechve study of ceftazldrme versus ceftazidime plus flucloxacillrn rn the emprnc treatment of febrrle episodes in se-
ON CIPROFLOXACIN
/BAYSTON
ET AL
verely neutropenrc patrents. Antimrcrob Agents Chemother 1985; 28: 824-828. 7. Wade JC, Schimpff SC, Newman KA, Wiermk PA: Stapbyiococcus epiderm,dIs: an increasing cause of infection in patients wrth granulocytopena Ann Intern Med 1982; 97: 503-508. 8. Wrnston DJ, Dudnrk DV, Chaprn M, et a/: Coagulase-negative staphylococcal bacteremra rn patients recerving Immunosuppressive therapy. Arch Intern Med 1983; 143: 32-36. 9. Cohen J, Donnelly JP, Worsley AM, Catovsky D, Goldman JM, Galton DAG, Septrcaemra caused by vrndans streptococci in neutropenrc patients with leukaemra Lancet 1983: II: 145221454. 10. Barrv AL. Jones RN: In v&o activitv of crorofloxacin against gram-oosltive COCCI.Am J Y I Med 1987; 82 (suppl 4A): 27-32. ’ 11. Smith GM, Leyland MJ, Farrell ID, Geddes AM: Prelrminary evaluation of ciprofloxac~n. a new 4-qulnolone antrbrotrc, rn the treatment of febrrle neutropenrc patrents. J Antrmrcrob Chemother 1986; 18 (suppl D): 165-174. 12. Wood ME, Newland AC: Intravenous crprofloxacrn in the treatment of Infection in rmmunocompromised patients. J Antrmrcrob Chemother 1986; 18 (suppl D): 175% 178. 13. Smith GM, Leyland MJ, Farrell ID, Geddes AM A clinrcal mrcrobiologrcal and pharmacokinetrc study of crprofloxacin plus vancomycin as rnrtral therapy of febrrle episodes in neutropenrc patrents. J Antrmrcrob Chemother 1988; 21: 647-655. 14. Klastersky J, Glauser MP. Schrmpff SC, et a/: Prospective randomrzed comparrson of three antrbrotic regimens for empirical therapy of suspected bacteremrc infection rn febrile granulocytopenic patients. Antimicrob Agents Chemother 1986; 29: 263-270.
November 30, 1989
The American Journal of Medicine
Volume 87 (suppl 5A)
5A-273S
F. BAYSTON, F.R.C.P.,’ SUEWANT, Ph.D,
JONATHAN COHEN,
This study developed further clinical experience in using a single agent (“monotherapy”) as empirical treatment for neutropenic patients with fever, and compared the safety and toxicity of two candidate agents, ceftazidime and ciprofloxatin. A prospective, randomized, single-center efficacy and safety comparison was conducted of intravenous ciprofloxacin, 200 mg every 12 hours, and ceftazidime, 2 g every eight hours, as initial empirical therapy in neutropenic patients with fever. Regimens were modified as necessary, guided by laboratory results and/or the clinical condition. Response was evaluated at 72 hours and at the end of the neutropenia. Toxicity was evaluated by regular clinical examination and laboratory investigations. A total of 43 patients with 51 febrile neutropenic episodes were enrolled into the study and randomly assigned to one of the two regimens. Five episodes were excluded from evaluation of efficacy because of protocol violations, leaving 46 evaluable episodes (21 ciprofloxacin, 25 ceftazidime). The two groups were well matched for risk factors for infection. There were no differences between the two groups in response rates either at 72 hours or at the end of neutropenia, although in the vast majority of patients some modification of the initial therapy was required. No patients died of uncontrolled bacterial infection. Superinfection with gram-positive cocci (often streptococci) was seen primarily in bone marrow transplant recipients who had been randomly assigned to receive ciprofloxacin. This study demonstrated that, in certain circumstances, a single antibiotic can be used successfully as initial empirical therapy in febrile neutropenic patients. In this study, ceftazidime and ciprofloxacin were generally of equal efficacy, but there appeared to be an increased incidence of streptococcal superinfection in bone marrow transplant recipients who received ciprofloxacin.
From the Infectious Diseases Unit. Departments of Bacteriology and Medicine, Hammersmith Hospital and Royal Postgraduate MedIcal School. London, United Kingdom. This work was supported by a grant from Bayer (UnIted Kingdom). Requests for reprints should be addressed to Dr. Jonathan Cohen. Hammersmith Hospital, Du Cane Road, London W12 ONN, Unlted Kingdom ‘Katherine F. Bavston is an Overseas Research Fellow of the MedIcal Research Council of New Zealand. tJonathan Cohen IS a Weilcome Trust Senior Lecturer.
of
MSc., F.R.C.P.t London, Umted Kfngdom
M
any clinical studies conducted over- the last 15 years have clearly established the benefit of the early empirical use of antibiotics in febrile neutropenic patients [1,21. The success of this approach is attributable, in part, to a knowledge of the epidemiology of infection in these patients, and, in part, to the growing number of potent, bactericidal antibiotics that have become available. Recent changes in both of these factors have led to a re-evaluation of the choice of antibiotics to be used for empirical therapy, notably, the suggestion that a single agent (so-called monotherapy) might be as effective as the traditional combination of a penicillin coupled with an aminoglycoside [3]. Ceftazidime is an extended spectrum cephalosporin that has been used successfully as a single agent for empirical therapy [4-61. Another. drug that merits consideration in this role is ciprofloxacin, a fluoroquinolone antibiotic with a favorable in v&o profile and the potential advantage of both oral and parenteral formulations. We have, therefore, carried out a prospective, randomized comparison of these two drugs as initial empirical therapy in a group of febrile neutropenic patients.
METHODS Entry Criteria This single-center trial was conducted at the Hammersmith Hospital, London, United Kingdom. All enrolled patients were receiving inpatie.nt treatment and met the following study criteria: age at least 18 years; presence of fever (defined as a temperature of 38°C on two consecutive occasions two hours apart or a single recording of 39°C or higher); and neutropenia (defined as a total granulocyte count of 1 x log/liter or less). Exclusi’on criteria included known allergy to a trial drug; pregnancy or lactation; serum creatinine level greater than 3 mgidl; suspected central nervous system infection; infection caused by an organism known to be resistant to either of the trial drugs; and, finally, the use of any effective antimicrobial therapy active against the infecting pathogen within 72 hours of trial entry, other than antibiotics used for selective alimentary tract contamination. No patient could enter the study more than once during the same episode of neutropenia.
TRIAL DESIGN AND TREATMENT DEFINITION Patient Allocation Patients were randomly allocated to one of the two treatment groups, using a computer-generated randomization. Treatment was not blinded.
November 30, 1989
The American Journal of Medicine
Volume 87 (suppl 5Aj
5A-269s
SYMPOSIUM
ON CIPROFLOXACIN
/ BAYSTON ET AL
Drug Treatment Patients with normal renal function received either ciprofloxacin 200 mg every 12 hours by intravenous infusion, or ceftazidime 2 g every eight hours as an intravenous bolus. Dosages for patients with impaired renal function were modified appropriately. Drug levels were not routinely monitored. All patients received gut decontamination from the time of admission to hospital. This generally consisted of co-trimoxazole 960 mg twice daily and oral nonabsorbable antifungal agents. Colistin 1.5 mU every six hours was added if resistant organisms were found in surveillance culture specimens. In addition, bone marrow transplantation patients received prophylactic acyclovir. Modification of therapy, defined as the use of any other antimicrobial agent, including the use of granulocyte transfusions, either in addition to or in substitution for the assigned drug regimen, could be made after study enrollment based on either clinical deterioration or documentation of antibiotic resistance. Duration of Therapy Study patients were first assessed at 72 hours after being randomly assigned to therapy. Patients who were judged to have had responses to treatment continued therapy with that agent until neutropenia had resolved or until they had been afebrile for at least four consecutive days and there was evidence of satisfactory control of focal infection. For those patients who had not had responses to treatment? therapy was altered as necessary and treatment with the study drug continued or discontinued at the discretion of the primary physician. Study therapy was stopped if an allergic reaction developed. All study patients were observed until resolution of infection and neutropenia, or death. Clinical and Laboratory Monitoring Prior to starting treatment, a complete history was obtained and a complete physical examination was performed on all patients. Patients were examined daily and the appearance of rashes or other possible allergic manifestations were noted. All patients had twice-weekly chest radiographic examinations from the time of admission, regardless of symptoms. Blood counts were performed daily and biochemistry profiles obtained at least on alternate days during neutropenia. Additional diagnostic tests were performed as required. Bacteriologic Evaluation Twice-weekly surveillance culture specimens of feces, urine, and throat gargles were obtained from all patients. Prior to commencing trial therapy, blood culture specimens and culture specimens of any suspected sites of infection were obtained. In patients with an indwelling venous catheter (most commonly a Hickman-Broviac) at least one sample was obtained through the catheter and, where possible, another from a peripheral vein. Blood culture specimens were obtained and paired samples were cultured aerobically and anaerobically using the Bactec system. Follow-up blood culture specimens were taken daily during persistent fever. Documented sites of infection were recultured whenever possible within 72 hours of clinical response and/or discontinuation of antibiotic therapy. Bacterial isolates were identified by standard bacte5A-270s
November 30, 1989
The American Journal of Medicine
riologic techniques. Antibiotic susceptibility was determined in the first instance by disk diffusion and subsequently by the microtiter technique for minimal inhibitory concentrations.
CLASSIFICATION OF INFECTION Febrile episodes were assigned to one of four groups. Episodes accompanied by definite clinical signs and symptoms of infection plus microbiologic confirmation were termed microbiologically documented infections, with or without bacteremia, based on blood culture specimen results. Urinary tract infections were diagnosed when a midstream urine specimen contained more than lo5 colony-forming units/ml. Episodes accompanied by definite clinical features of infection, but without microbiologic proof, were termed clinically documented infections. Pneumonia was diagnosed by the presence of a new pulmonary infiltrat,e not attributable to a noninfectious cause. All other febrile episodes were classified as either possible or doubtful infections. Superinfection was taken to be the isolation of an organism different from that initially causing fever (or the first isolation of an organism after a period of response) during the course of therapy. Evaluation of Response Response was evaluated at 72 hours and again at the end of therapy. At the first evaluation, outcome was classified in one of three categories: (1) success, defined as a return of temperature to normal (i.e., temperature less than 37.5”C), and maintenance of an afebrile state for at least 48 hours thereafter; (2) improvement, defined as temperature less than 38°C and clinically improved without modification of trial therapy; or (3) failure! defined as death from infection or clinical deterioration necessitating alteration of therapy. At the second evaluation, episodes were classified as: (1) success, defined as an episode in which recovery from infection and neutropenia was achieved without modification of therapy; (2) success with modification, defined as the patient recovering from an episode of infection but requiring modification of the assigned regimen; and (3J failure, defined as death occurring during the febrile neutropenic episode. Toxicity Renal or hepatic abnormalities were considered to be related to the study drugs in the absence of other attributable causes. Patients in whom features of septic shock developed, who received cyclosporin A or amphotericin B, or in whom hepatic veno-occlusive disease or graft-versus-host disease involving the liver developed during the period of therapy were excluded from assessment of toxicity. Statistics Fisher’s exact test was used to compare the distribution of febrile episodes and the outcome of therapy, and the Student t test was used to compare duration of neutropenia in the two study groups.
RESULTS Characteristics of the Study Population Over a 24-month period, 43 patients with 51 neutropenic febrile episodes were randomly assigned to receive single-agent empirical therapy with either ciprofloxacin or ceftazidime. Five episodes were excluded
Volume 87 (suppl 5A)
SYMPOSIUM
from evaluation because of trial violations, leaving 46 evaluable episodes in 40 patients. Of these, 21 were treated with ciprofloxacin and 25 with ceftazidime. As shown in Table I, the patients in each group were comparable with respect to age, sex, duration of neutropenia, and underlying diagnosis. Neutropenic episodes generally occurred in the context of bone marrow transplantation for chronic myeloid leukemia or acute myeloid leukemia (57 percent overall), or chemotherapy for acute leukemia (41 percent). An additional patient in the ciprofloxacin group became neutropenic after therapy for blast transformation of chronic myeloid leukemia. Classification of Episodes There were no statistically sigtiificant differences in the distribution of febrile episodes between the two groups. Microbiologically documented episodes occurred in seven of 21 episodes (33 percent) in the ciprofloxacin group, all of which were baeteremias, and in 16 of 25 (64 percent) treated episodes in the ceftazidime group (p = 0.07). Of the latter, 11 were solely bacteremic episodes, four were urinaky tract infections alone, and, in one patient, bacteremia and urinary tract infection (caused by different organisms) occurred together. The remaining episodes in both groups were classified as clinically documented episodes. Outcome Evaluatian There was no difference in time to defervescence between the two groups (ciprofloxacin, 5.05 + 4.18 days; ceftazidime, 4.6 i 3.34 days; p = 0.7). At 72 hours, the outcome for episodes treated with ciprofloxacin was no different from that for episodes treated with ceftazidime. Success occurred in seven of 21 (33 percent) versus 10 of 25 (40 percent), improvement in three of 21 (14 percent) versus three of 25 (12 percent), and failure in 11 of 21 (52 percent) versus 12 of 25 (48 percent) recipients of ciprofloxacin versus ceftazidime, respectively. No patients died during this period. At the end of therapy, outcome was classified as success with modification in the majority of instances: 15 6f 21 (71 percent) episodes treated with ciprofloxatin and 16 of 25 (64 percent) episodes treated with ceftazidime (not significant). Success without modification was documented in only three of 21 (14 percent) ciprofloxacin patients and seven of 25 (28 percent) ceftazidime patients, and failure was documented in three of 21 (14 percent) and two of 25 (8 percent) patients, respectively. Details of bacterial isolates for each treatment group are given in Table II and sensitivity data (minimal inhibitpry concentrations) in Table III. There was no difference in outcome for microbiologically documented infections between the two treatment groups, either at 72 hours (p = 0.3) or at the end of therapy (p = 0.21). Bacterial .superinfections (six patients) occurred predominantly in bone marrow transplant recipients (five of si?i patients) and most commonly in recipients of ciprofloxacin: four streptococcal bacteremiai and one Capnocytophaga ochracea bacteremia. In the ceftazidime group, there was one episode of ClostAdimn d@Ele-associated diarrhea. Fungal superinfection occurred in one patient in each treatment group; both were bone marrow transplant recipients. November
’ TABLE
ON CIPROFLOXACINiBAYSTON
ET AL
I
Characteristics
of the Study
Population* Treatment Group
Characteristic
Ciprofloxacin
Number Sex (M:F] Ag;~;;rsl
Ceftazidime
7:14
Range Neutropenia Mean days from trial entry SD Dlagnosls BMP Acute leukemia Other
I&
256
264
18.86 13.39
17.88 9.24
12 8 1
if 0
BMT = bone marrow transplant: SD = standard deviation. *Evaluable patients only.
TABLE
II
Organisms
Isolated
in Primary
Infections* Treatment Group Ciprofloxacin
Ceftazidime
Single organisms Staphylococcus epidermldis
Streptococcus sanguis Streptococcus mitis Colynebacterium sp. C. ochracea Eschenchia cali E.co/d Pseudomonas aeruginosa Mlxed infections
S epiderm~d,siAcnetobactersa. S. epiderm/disiS. sangua/Streptococcus equrnus Corynebacterium sp.lE co/it
0
*Organisms are bacteremic isolates unless otherwe kkinary tract isolates.
TABLE
0 1 1
h stated
Ill
Sensitivities Infections
for All Bacterial and Superinfections
Isolates,
Including
Primary
MIC (fig/ml)* Organism S. epidermidis S.sanguis S.mitis Streptococcus equinus Streptococcus pneumonfae Corvnebacterium SD.
/
Number i 3 i 1
Ciprofloxacin
Ceftazidime
0.15 (0.06-0.25)t
23 [3.0-501 1.2 (0.05-3.0) 0.1 0.1
1.7 ‘;:y”] 0.01
4.0 ND
E CO//
8
P aeruginosa Acinetobacter sp. C ochracea
i
0.0;
2
8.0
0.01 (<.y.Ol)
iii (0.05$01)
i:;
MIC = minimal lnhlbitory concentration; ND = no data. *MinImal inhibitory concentrations are expressed as means. tNumbers In parentheses, ranges.
Skin rash possibly attributable to administration of a study antibiotic was seen in two patients from each group. Two patients, one from each study group, developed pneumonitis in addition to a skin rash. These events occurred in bone marrow transplant patients 30, 1989
The American Journal of Medicine
Volume 87 (suppl 5A)
5A-271s
SYMPOSIUM
ON CIPROFLOXACIN
/ BAYSTON ET AL
,ipe~g;,iP$”Ceftazldime +-----4
Vancomycln Amphoterun 1-1 Ciprofloxacin 1-1
+364,,
0
1
2
3
4
/
5
/I
6 7 Days
‘,
8
H
9
1,
IO
11
12
1
13
Figure 1. Temperature chart of a bone marrow transplant patient randomly assigned to receive ciprofloxacin. Antibiotics used are Indicated at the top of the chart. Bars, duration of antibiotic administration.
and were probably unrelated to antibiotic therapy. No hematologic or biochemical toxicity could be directly attributable to either of the study drugs. All but one death during the period of neutropenia occurred in bone marrow transplant patients. None of these patients had documented bacterial infection at the time of death. In the ciprofloxacin-treated group, two patients died with idiopathic pneumonitis and one with Aspergillus pneumonia. Of patients treated with ceftazidime, one died with an idiopathic pneumonitis and the other patient developed influenza B pneumonia following graft rejection and died of multi-organ failure.
COMMENTS These results provide further support for the view that a single antibiotic can be used safely as initial empirical therapy in febrile neutropenic patients. Five of the 46 evaluable patients died during neutropenic episodes, but in none was this due to documented bacterial infection. Thus, taking death due to bacterial infection as the ultimate outcome, ceftazidime and ciproflox&in were equally effective. Furthermore, no differences were apparent when comparing the frequency with which it was necessary to add to, or modify, the initial regimen. Seventy-two hours after starting therapy, approximately half the patients in each group (ciprofloxacin, 47 percent; ceftazidime, 52 percent) had responses such that no change of therapy was required. Taken with the remaining patients who did require alteration of their treatment, overall survival at the end of neutropenia was 18 of 21 (86 percent) patients in the ciprofloxacin-treated group and 23 of 25 (92 percent) patients’ in the ceftazidimetreated group. These results are broadly comparable with previous experience with ceftazidime monotherapy, which documented an initial response rate of 50 to 60 percent and an overall success rate of more than 95 percent [4,51. Bowever, a more detailed analysis of the results revealed one important difference between the two drugs in relation to the frequency of superinfections. These occurred almost exclusively in the subset of bone marrow transplant recipients (seven of 26 versus one of 20 patients). Four of the six episodes due to bacteria were caused by streptococci, and all four were in the ciprofloxacin group. The growing significance of gram-positive bacteremia as a complication of neutropenia has been exten-
5A-2725
November 30, 1989
The American Journal of Medune
sively documented [7,8]. The single most common group of isolates have been coagulase-negative staphylococci, but we and others have also drawn attention to the importance of Streptococcus viridans [9]. These bacteremias most likely originate in the oral cavity, and one may speculate that their predominance in the bone marrow transplant recipients reflects the damage to the oral mucosa caused by the more intense conditioning regimens used in these patients. It is of interest that streptococcal superinfection was a particular problem in the patients receiving ciprofloxacin, reflecting that drug’s modest i,n vitro activity against streptococci [lo]. Indeed, previous studies with ciprofloxacin in neutropenic patients have provided information on the development of resistance to ciprofloxatin among gram-positive cocci [ll!, prompting its use in combination with benzylpenicilhn [12] or vancomytin [131 to provide additional therapy. Our data confirm that it is frequently necessary to add anti-streptococcal or anti-staphylococcal antibiotics, but suggest that this can be guided by the clinical circumstances; not all patients need vancomycin. In addition, these data illustrate well the importance of careful definition of terms. If one takes the view of the European Organization for Research on Treatment of Cancer (EORTC) that the purpose of empirical therapy is to achieve “. . . a lasting return of temperature to the normal level . . . and resolution of all signs and symptoms, without addition of other antibiotics” [14], then an initial response rate of 30 to 40 percent at the first evaluation is clearly unacceptable. An alternative view [5] is that empirical therapy is designed only to prevent death in the early stages before a more specific, “tailored” antibiotic regimen is chosen based on bacteriologic results and the patients’ initial response. On this basis, the only cause of failure would be death due to infection during the period of neutropenia, yielding an overall success rate for this study of approximately 90 percent. Despite the apparently compelling logic of this reasoning, it can lead to problems of interpretation. Consider the clinical course of the patient illustrated in Figure 1, who was randomly assigned to receive ciprofloxacin but in whom multiple modifications of therapy were later required before a satisfactory clinical response was obtained. Using the criteria developed by Pizzo et al [51, we would be obliged to classify this patient as “success with modification,” yet the implication that the empirical use of ciprofloxacin had contributed to the successful outcome is difficult to justify and potentially misleading. On the other hand, the more rigorous criteria employed by the EORTC provides a clearer basis for comparison between different agents, but could hinder the assessment of novel approaches that do not lend themselves to this type of evaluation. These problems of definition should not be allowed to deflect attention away from the more important issue of the potential value of monotherapy. The results presented here add weight to the growing body of evidence that suggests that in selected circumstances a single drug can be used as empirical therapy for fever in neutropenic patients but underlines the importance of distinguishing carefully between risk groups within the general category of neutropenia. Both ciprofloxacin and ceftazidime are potential candidates, and merit further study.
Volume 87 (suppl 5A)
SYMPOSIUM
ACKNOWLEDGMENT We are grateful to the staff of the Haematology Department, Hammersmith Hospital, for allowing us to study patients under their care. REFERENCES 1. Schlmpff SC, Satterlee W, Young VM, Serprck A: Emprric therapy with carbenrcrll~n and gentamrcrn for febrile patrents wrth cancer and granulocytopenra. N Engl J Med 1971; 284: 1061-1065. 2. Schimpff SC: Emprric anhbrotic therapy for granulocytopenrc cancer patrents. Am J Med 1986; 80 (suppl 5C): 13-20. 3. Hathorn JW, PIZZO PA: Is there a role for monotherapy wrth beta-lactam antibiotrcs rn the inihal emoirical manaeement of febrile neutrooenic cancer oatrents? J Antrmrcrob Chemother 1986; 17 (sup;1 A): 41-54. 4. Donnelly JP, Marcus RE, Goldman JM, et a/: Ceftazldime as first-lrne therapy for fever in acute leukaemra. J Infect 1985; 11. 205-215. 5. Pizza PA, Hathorn JW, Hiemenz J, era/: A randomrzed trial comparrng ceftazidime alone wrth combrnation antibiotic therapy in cancer patients with fever and neutropenia. N Engl J Med 1986; 315, 552-558. 6. De Pauw B, Wrlliams K, De Neeff J. eta/: A randomrzed prospechve study of ceftazldrme versus ceftazidime plus flucloxacillrn rn the emprnc treatment of febrrle episodes in se-
ON CIPROFLOXACIN
/BAYSTON
ET AL
verely neutropenrc patrents. Antimrcrob Agents Chemother 1985; 28: 824-828. 7. Wade JC, Schimpff SC, Newman KA, Wiermk PA: Stapbyiococcus epiderm,dIs: an increasing cause of infection in patients wrth granulocytopena Ann Intern Med 1982; 97: 503-508. 8. Wrnston DJ, Dudnrk DV, Chaprn M, et a/: Coagulase-negative staphylococcal bacteremra rn patients recerving Immunosuppressive therapy. Arch Intern Med 1983; 143: 32-36. 9. Cohen J, Donnelly JP, Worsley AM, Catovsky D, Goldman JM, Galton DAG, Septrcaemra caused by vrndans streptococci in neutropenrc patients with leukaemra Lancet 1983: II: 145221454. 10. Barrv AL. Jones RN: In v&o activitv of crorofloxacin against gram-oosltive COCCI.Am J Y I Med 1987; 82 (suppl 4A): 27-32. ’ 11. Smith GM, Leyland MJ, Farrell ID, Geddes AM: Prelrminary evaluation of ciprofloxac~n. a new 4-qulnolone antrbrotrc, rn the treatment of febrrle neutropenrc patrents. J Antrmrcrob Chemother 1986; 18 (suppl D): 165-174. 12. Wood ME, Newland AC: Intravenous crprofloxacrn in the treatment of Infection in rmmunocompromised patients. J Antrmrcrob Chemother 1986; 18 (suppl D): 175% 178. 13. Smith GM, Leyland MJ, Farrell ID, Geddes AM A clinrcal mrcrobiologrcal and pharmacokinetrc study of crprofloxacin plus vancomycin as rnrtral therapy of febrrle episodes in neutropenrc patrents. J Antrmrcrob Chemother 1988; 21: 647-655. 14. Klastersky J, Glauser MP. Schrmpff SC, et a/: Prospective randomrzed comparrson of three antrbrotic regimens for empirical therapy of suspected bacteremrc infection rn febrile granulocytopenic patients. Antimicrob Agents Chemother 1986; 29: 263-270.
November 30, 1989
The American Journal of Medicine
Volume 87 (suppl 5A)
5A-273S