Journal of Antimicrobial Chemotherapy (1992) 30, 843-854
A comparative study of imipenem versus piperadllin pins gentamidn in the initial management of febrile neotropenic patients with haematological malignancies M. J. Leyland*, K. F. Bayston*, J. Cohen*, R. Warren', A. C. Newtamf, A. J. Blnf, C. Cefar, D. G. Whit*/, S. A. Marra/, D. BarefonP and N. B. Deaney*
Three-hundred and twelve episodes of fever in 234 neutropenic patients with haematological malignancies were treated empirically with either imipenem or a combination of piperadllin and gentamicin. There were no significant differences in the percentages of patients responding to therapy at either 72 h (59% and 56% of assessable episodes in the imipenem and combination groups respectively) or at the end of treatment (55% and 53% of assessable episodes in the imipenem and combination groups respectively). Patients in the piperaciUin plus gentamicin group experienced significantly more renal tubular damage whereas those who received imipenem suffered more nausea or vomiting. We conclude that imipenem monotherapy represents an acceptable alternative to piperaciUin plus gentamicin as empirical therapy of the febrile neutropenic patient
Introduction The necessity for the early initiation of empirical antimicrobial chemotherapy in the treatment of febrile neutropenic patients with haematological malignancies is wellestablished (Schimpflf er al., 1971). A combination of bactericidal antibiotics is usually used in order to broaden the spectrum of activity and to exploit the potential for synergy (Klastersky & Zinner, 1982). Widely used combinations have consisted of a ureidopenicillin and an aminoglycoside such as amikacin (Norrby et al., 1987) or gentamicin (Boughton et al., 1989). However, because of the nephro- and ototoxicity associated with the aminoglycosides, the efficacies of double /Mactam combinations (Clough et al., 1985; Young, 1985) and, more recently, of monotherapy (Hathorn, Rubin & Pizzo, 1987) have been assessed. Early studies of single-agent therapy with carboxypenicillins (Bodey et al., 1971) and the ureidopenicillins (Wade et al., 1981) were disappointing because of poor response rates and the emergence of resistant strains. When newer, broad spectrum agents such as ceftazidime and ciprofloxacin 843 0305-7453/92/120843+12 $08.00/0
© 1992 Tne British Society for Antimicrobial Chemotherapy
Downloaded from http://jac.oxfordjournals.org/ at :: on January 7, 2015
'Department of Haematology, East Birmingham Hospital, Birmingham, B9 5ST; Infectious Diseases Unit, Departments of Bacteriology and Medicine, Hammersmith Hospital, London; 'Department of Microbiology, New Addenbrookes Hospital, Cambridge; ^Department of Haematology, London Hospital, Whitechapel, London; 'Department of Microbiology, Royal Victoria Infirmary, Newcastle upon Tyne, NE1 4LP; ^Public Health Laboratory, Royal United Hospital, Bath; 'Department of Haematology, Dudley Road Hospital, Birmingham, B18 7QH;kMedical Department, Merck Sharp A. Dohme Ltd., Hertford Road, Hoddesdon, UK b
844
M J. Leylmod et aL
Patients and methods Patients Patients were recruited in seven centres in England over a period of 20 months. Ethics Committee approval was obtained in each centre. Neutropenic patients with haematological malignancies who were at least 18 years of age and who became pyrexial were entered into the study. Neutropenia was defined as a granulocyte count of less than 1-0 x 10*/L and fever as two temperature readings of > 38°C 2 h apart or one reading of > 39°C. Patients were not entered into the study more than once during a single neutropenic episode. Other exclusions included an allergy to any of the test antibiotics and an infection caused by an organism known to be resistant to any of the trial drugs. Patients who were receiving antimicrobial agents for selective gut decontamination were not excluded. All patients gave their consent to participate.
Investigations All patients were examined at entry in an attempt to identify the site of infection. Routine haematological investigations were carried out daily and tests of renal and hepatic function were performed before starting therapy, during the course and on completion. Paired aerobic and anaerobic blood cultures were taken from all patients and from a Hickman catheter if present. A mid-stream urine (MSU) as well as swabs from any other clinically relevant sites were also cultured. Repeat blood cultures were obtained in patients with persistent fever ( > 37-5°Q and subsequent urine specimens were collected from patients with documented urinary tract infection during and one week following completion of therapy.
Downloaded from http://jac.oxfordjournals.org/ at :: on January 7, 2015
became available it was thought that they might represent alternatives to the pcnicillinaminoglycoside combinations (Pizzo et al., 1986; Verhagan et al., 1987; Smith et al., 1988; Liang et al., 1990). Imipenem has also been used successfully as monotherapy (Norrby et al., 1987; Mortimer et al., 1988; Liang et al., 1990). In recent years the incidence of infections caused by Gram-positive bacteria has been increasing in this patient population (Wade et al., 1982). Compared with their potential anti-Gram-negative activities, ceftazidime and ciprofloxacin are less effective in vitro against Gram-positive organisms and this has led some investigators to suggest that glycopeptides should routinely be combined with these agents; this of course eliminates the advantages of monotherapy (de Pauw, 1990). Although some Gram-positive bacteria are susceptible in vitro to the ureidopenicillin-aminoglycosidc combinations, many of the coagulase-negative staphylococcal isolates are resistant. Imipenem possesses the increased anti-Gram-negative activity of the newer agents as well as enhanced activity against Gram-positive bacteria, particularly staphylococci (Kropp et al., 1980). It also has useful activity against some strains of methicillin-resistant Staphylococcus catreus (Cherubin et al., 1981; Reeves, Bywater & Holt, 1987) and Staphylococcus epidermidis (Livingston, Elliott & Cobbs, 1981). We described a multicentre, randomized, controlled clinical trial in which imipenem was compared with the ureidopenicillin-aminoglycoside combination used most widely in the UK. for the treatment of febrile granulocytopenic patients with haematologjcal malignancies.
ImJpenem in the febrile neotropenfc patient
845
Antibiotics
Assessment The response to therapy was assessed retrospectively by a panel of four investigators (J.C., MJ.L., R.W. and D.G.W.), initially at 72 h and again on completion of therapy. The assessment was made without knowledge of the treatment group to which the patient had been assigned. Strict criteria were observed for assigning patients to each outcome category (Figure). At 72 h patients were considered to have responded or not Assessment (final) on completion of therapy Assessment at 72h
Cured « 3 8 * C )
Responded ( S 37-5*C ond maintained for 48h)
/ , Entry
Failed (23B*C, relapse or reinfection)
X
/
/
Not responded O 3 7 - V C or mointoJnGd tec 4 8 h) \ \
Unassessablt (other onlimicrotriali added empirically)
Cured (reduction of temperature < 3 7 - 5 * C subsequently maintained)
Failed (persistent pyrexio, reinfection or relapse)
Unevaluable (insuffidtnt durotloo of no data)
Flgmc Flow diagram for the assessment of clinical outcome.
Unassessobte (other antimicrobials added empirically)
Downloaded from http://jac.oxfordjournals.org/ at :: on January 7, 2015
Patients were randomly allocated to receive either imipenem or a combination of piperaciUin and gentamicin by opening sealed envelopes. The dosages varied according to body weight, renal function and severity of infection. The median dose if imipenem was 3-5 g/day (10th and 90th deciles, 2 and 4 g respectively) administered intravenously in four divided doses; the median dose of piperaciUin was 16 g/day (10th and 90th deciles, 16 and 16 g respectively) administered intravenously in four divided doses; the median starting dose of gentamicin was 240 mg/day (10th and 90th deciles, 200 and 360 mg respectively) given in three divided doses. The dosage was subsequently adjusted in the light of serum concentrations (median trough concentration 1-S mg/L; median peak concentration 5-8 mg/L in evaluable patients). Antibiotics were continued until the patient was afebrile or until they were considered inappropriate; imipenem was administered for a median duration of seven days (10th and 90th deciles, three and 12 days respectively) and the combination for a median of seven days (10th and 90th deciles, four and 13 days respectively). The decision to use selective gut decontamination was left to each centre. Anti-viral and anti-fungal agents were given as necessary.
846
M. J. LcyUnd et aL
according to their temperature response, or were deemed unassessable for one of the following reasons: the addition of empirical, non-trial antimicrobials, missing temperature data, insufficient duration of treatment ( < 72 h in the absence of clinical deterioration) or the isolation of a bacterium resistant to the assigned therapy. On completion of therapy patients were defined as cured or failed, depending on their temperature response, or were unassessable. The administration of antimicrobial agents (including antifungals and antivirals) to the initial treatment regimen was a cause of failure. Patients who had not responded at 72 h could still be cured by the final assessment. However, patients who were unassessable at 72 h continued to be unassessable. In addition, each febrile episode was classified as follows (EORTC, 1978): (a) Microbiologically documented infection (with or without bacteraemia); (b) Clinically documented infection; no pathogen identified; (c) Possible infection; pathogen and clinical site not identified; (d) Infection dubious; fever retrospectively thought not to be due to infection.
Jn-vitro studies Significant bacterial isolates were identified using conventional laboratory procedures and their susceptibilities to imipenem, gcntamicin, piperacillin and other antibiotics were determined by a disc diffusion method (Stokes & Waterworth, 1972). Statistics Comparisons of the major outcome variables were made by applying 95% confidence intervals to the differences between the treatment groups. Other differences were analysed by the Chi-square test, unless stated otherwise in the text. Results There were 312 episodes of fever in the 234 patients who were entered into the study. As shown in Table I, 35 of the 164 episodes in patients assigned to the imipenem regimen were not evaluable for efficacy because of protocol violations, compared with 25 episodes in patients receiving combination therapy. In addition to these protocol violations a number of patients were not assessable at one or the other of the assessment times for a variety of reasons, including lack of sufficient data or the empirical administration of non-trial antimicrobial agents. The patients in both treatment groups were similar in terms of age, sex and underlying haematological diseases, although there was a slight excess of patients with Hodgkin's disease in the imipenem group. The neutrophil counts and the durations of neutropenia before starting therapy were comparable for both groups (Table II). Differences in the numbers of patients in each treatment group receiving absorbable (co-trimoxazole) or non-absorbable (colistin and/or framycetin) prophylactic antibiotics were not significant. The classification of the 252 evaluable febrile episodes is shown in Table m .
Downloaded from http://jac.oxfordjournals.org/ at :: on January 7, 2015
Patient tolerance of the treatment regimens was evaluated according to the presence or absence of nausea or vomiting, pain, erythema or exudate at the infusion site and/or phlebitis of the cannulated vein. All adverse reactions which were probably or definitely related to the therapeutic regimens, including those identified by haematological and biochemical investigations, were also recorded.
Imipenem In tbe febrOe neatropenk patient
847
Table L Episodes of fever excluded from evaluation
No. of episodes treated No. (%) of protocol violations Non-haematological malignancy Incorrect dose of study drug Other antibiotics given Patient not neutropenic Patient not pyrexial Other violations No. of assessable episodes
imipenem
Treatment group combination
total
164 35(21%) 5 0 13 13 4 5
148 25 (17%) 6 2 7 8 6 1
312 60 (19%) 11 2 20 21 10 6
129
123
252
At 72 h 68 (59%) of the evaluable episodes treated with imipenem had responded compared with 65 (56%) of those treated with piperacillin and gentamicin. There were 13 and six unassessable episodes in the imipenem and combination groups, respectively. At the end-of-treatment evaluation 58 (55%) of the patients in the imipenem group were assessed as cured compared with 58 (53%) of those in the gentamicin-piperacillin group (Table IV). Four of the patient episodes in the monotherapy group and two in Table n . Demographic and clinical characteristics of the evaluable patient episodes Characteristic No. of evaluable episodes Median age (range) % of patients with neutrophil count of entry
A comparative study of imipenem versus piperadllin pins gentamidn in the initial management of febrile neotropenic patients with haematological malignancies M. J. Leyland*, K. F. Bayston*, J. Cohen*, R. Warren', A. C. Newtamf, A. J. Blnf, C. Cefar, D. G. Whit*/, S. A. Marra/, D. BarefonP and N. B. Deaney*
Three-hundred and twelve episodes of fever in 234 neutropenic patients with haematological malignancies were treated empirically with either imipenem or a combination of piperadllin and gentamicin. There were no significant differences in the percentages of patients responding to therapy at either 72 h (59% and 56% of assessable episodes in the imipenem and combination groups respectively) or at the end of treatment (55% and 53% of assessable episodes in the imipenem and combination groups respectively). Patients in the piperaciUin plus gentamicin group experienced significantly more renal tubular damage whereas those who received imipenem suffered more nausea or vomiting. We conclude that imipenem monotherapy represents an acceptable alternative to piperaciUin plus gentamicin as empirical therapy of the febrile neutropenic patient
Introduction The necessity for the early initiation of empirical antimicrobial chemotherapy in the treatment of febrile neutropenic patients with haematological malignancies is wellestablished (Schimpflf er al., 1971). A combination of bactericidal antibiotics is usually used in order to broaden the spectrum of activity and to exploit the potential for synergy (Klastersky & Zinner, 1982). Widely used combinations have consisted of a ureidopenicillin and an aminoglycoside such as amikacin (Norrby et al., 1987) or gentamicin (Boughton et al., 1989). However, because of the nephro- and ototoxicity associated with the aminoglycosides, the efficacies of double /Mactam combinations (Clough et al., 1985; Young, 1985) and, more recently, of monotherapy (Hathorn, Rubin & Pizzo, 1987) have been assessed. Early studies of single-agent therapy with carboxypenicillins (Bodey et al., 1971) and the ureidopenicillins (Wade et al., 1981) were disappointing because of poor response rates and the emergence of resistant strains. When newer, broad spectrum agents such as ceftazidime and ciprofloxacin 843 0305-7453/92/120843+12 $08.00/0
© 1992 Tne British Society for Antimicrobial Chemotherapy
Downloaded from http://jac.oxfordjournals.org/ at :: on January 7, 2015
'Department of Haematology, East Birmingham Hospital, Birmingham, B9 5ST; Infectious Diseases Unit, Departments of Bacteriology and Medicine, Hammersmith Hospital, London; 'Department of Microbiology, New Addenbrookes Hospital, Cambridge; ^Department of Haematology, London Hospital, Whitechapel, London; 'Department of Microbiology, Royal Victoria Infirmary, Newcastle upon Tyne, NE1 4LP; ^Public Health Laboratory, Royal United Hospital, Bath; 'Department of Haematology, Dudley Road Hospital, Birmingham, B18 7QH;kMedical Department, Merck Sharp A. Dohme Ltd., Hertford Road, Hoddesdon, UK b
844
M J. Leylmod et aL
Patients and methods Patients Patients were recruited in seven centres in England over a period of 20 months. Ethics Committee approval was obtained in each centre. Neutropenic patients with haematological malignancies who were at least 18 years of age and who became pyrexial were entered into the study. Neutropenia was defined as a granulocyte count of less than 1-0 x 10*/L and fever as two temperature readings of > 38°C 2 h apart or one reading of > 39°C. Patients were not entered into the study more than once during a single neutropenic episode. Other exclusions included an allergy to any of the test antibiotics and an infection caused by an organism known to be resistant to any of the trial drugs. Patients who were receiving antimicrobial agents for selective gut decontamination were not excluded. All patients gave their consent to participate.
Investigations All patients were examined at entry in an attempt to identify the site of infection. Routine haematological investigations were carried out daily and tests of renal and hepatic function were performed before starting therapy, during the course and on completion. Paired aerobic and anaerobic blood cultures were taken from all patients and from a Hickman catheter if present. A mid-stream urine (MSU) as well as swabs from any other clinically relevant sites were also cultured. Repeat blood cultures were obtained in patients with persistent fever ( > 37-5°Q and subsequent urine specimens were collected from patients with documented urinary tract infection during and one week following completion of therapy.
Downloaded from http://jac.oxfordjournals.org/ at :: on January 7, 2015
became available it was thought that they might represent alternatives to the pcnicillinaminoglycoside combinations (Pizzo et al., 1986; Verhagan et al., 1987; Smith et al., 1988; Liang et al., 1990). Imipenem has also been used successfully as monotherapy (Norrby et al., 1987; Mortimer et al., 1988; Liang et al., 1990). In recent years the incidence of infections caused by Gram-positive bacteria has been increasing in this patient population (Wade et al., 1982). Compared with their potential anti-Gram-negative activities, ceftazidime and ciprofloxacin are less effective in vitro against Gram-positive organisms and this has led some investigators to suggest that glycopeptides should routinely be combined with these agents; this of course eliminates the advantages of monotherapy (de Pauw, 1990). Although some Gram-positive bacteria are susceptible in vitro to the ureidopenicillin-aminoglycosidc combinations, many of the coagulase-negative staphylococcal isolates are resistant. Imipenem possesses the increased anti-Gram-negative activity of the newer agents as well as enhanced activity against Gram-positive bacteria, particularly staphylococci (Kropp et al., 1980). It also has useful activity against some strains of methicillin-resistant Staphylococcus catreus (Cherubin et al., 1981; Reeves, Bywater & Holt, 1987) and Staphylococcus epidermidis (Livingston, Elliott & Cobbs, 1981). We described a multicentre, randomized, controlled clinical trial in which imipenem was compared with the ureidopenicillin-aminoglycoside combination used most widely in the UK. for the treatment of febrile granulocytopenic patients with haematologjcal malignancies.
ImJpenem in the febrile neotropenfc patient
845
Antibiotics
Assessment The response to therapy was assessed retrospectively by a panel of four investigators (J.C., MJ.L., R.W. and D.G.W.), initially at 72 h and again on completion of therapy. The assessment was made without knowledge of the treatment group to which the patient had been assigned. Strict criteria were observed for assigning patients to each outcome category (Figure). At 72 h patients were considered to have responded or not Assessment (final) on completion of therapy Assessment at 72h
Cured « 3 8 * C )
Responded ( S 37-5*C ond maintained for 48h)
/ , Entry
Failed (23B*C, relapse or reinfection)
X
/
/
Not responded O 3 7 - V C or mointoJnGd tec 4 8 h) \ \
Unassessablt (other onlimicrotriali added empirically)
Cured (reduction of temperature < 3 7 - 5 * C subsequently maintained)
Failed (persistent pyrexio, reinfection or relapse)
Unevaluable (insuffidtnt durotloo of no data)
Flgmc Flow diagram for the assessment of clinical outcome.
Unassessobte (other antimicrobials added empirically)
Downloaded from http://jac.oxfordjournals.org/ at :: on January 7, 2015
Patients were randomly allocated to receive either imipenem or a combination of piperaciUin and gentamicin by opening sealed envelopes. The dosages varied according to body weight, renal function and severity of infection. The median dose if imipenem was 3-5 g/day (10th and 90th deciles, 2 and 4 g respectively) administered intravenously in four divided doses; the median dose of piperaciUin was 16 g/day (10th and 90th deciles, 16 and 16 g respectively) administered intravenously in four divided doses; the median starting dose of gentamicin was 240 mg/day (10th and 90th deciles, 200 and 360 mg respectively) given in three divided doses. The dosage was subsequently adjusted in the light of serum concentrations (median trough concentration 1-S mg/L; median peak concentration 5-8 mg/L in evaluable patients). Antibiotics were continued until the patient was afebrile or until they were considered inappropriate; imipenem was administered for a median duration of seven days (10th and 90th deciles, three and 12 days respectively) and the combination for a median of seven days (10th and 90th deciles, four and 13 days respectively). The decision to use selective gut decontamination was left to each centre. Anti-viral and anti-fungal agents were given as necessary.
846
M. J. LcyUnd et aL
according to their temperature response, or were deemed unassessable for one of the following reasons: the addition of empirical, non-trial antimicrobials, missing temperature data, insufficient duration of treatment ( < 72 h in the absence of clinical deterioration) or the isolation of a bacterium resistant to the assigned therapy. On completion of therapy patients were defined as cured or failed, depending on their temperature response, or were unassessable. The administration of antimicrobial agents (including antifungals and antivirals) to the initial treatment regimen was a cause of failure. Patients who had not responded at 72 h could still be cured by the final assessment. However, patients who were unassessable at 72 h continued to be unassessable. In addition, each febrile episode was classified as follows (EORTC, 1978): (a) Microbiologically documented infection (with or without bacteraemia); (b) Clinically documented infection; no pathogen identified; (c) Possible infection; pathogen and clinical site not identified; (d) Infection dubious; fever retrospectively thought not to be due to infection.
Jn-vitro studies Significant bacterial isolates were identified using conventional laboratory procedures and their susceptibilities to imipenem, gcntamicin, piperacillin and other antibiotics were determined by a disc diffusion method (Stokes & Waterworth, 1972). Statistics Comparisons of the major outcome variables were made by applying 95% confidence intervals to the differences between the treatment groups. Other differences were analysed by the Chi-square test, unless stated otherwise in the text. Results There were 312 episodes of fever in the 234 patients who were entered into the study. As shown in Table I, 35 of the 164 episodes in patients assigned to the imipenem regimen were not evaluable for efficacy because of protocol violations, compared with 25 episodes in patients receiving combination therapy. In addition to these protocol violations a number of patients were not assessable at one or the other of the assessment times for a variety of reasons, including lack of sufficient data or the empirical administration of non-trial antimicrobial agents. The patients in both treatment groups were similar in terms of age, sex and underlying haematological diseases, although there was a slight excess of patients with Hodgkin's disease in the imipenem group. The neutrophil counts and the durations of neutropenia before starting therapy were comparable for both groups (Table II). Differences in the numbers of patients in each treatment group receiving absorbable (co-trimoxazole) or non-absorbable (colistin and/or framycetin) prophylactic antibiotics were not significant. The classification of the 252 evaluable febrile episodes is shown in Table m .
Downloaded from http://jac.oxfordjournals.org/ at :: on January 7, 2015
Patient tolerance of the treatment regimens was evaluated according to the presence or absence of nausea or vomiting, pain, erythema or exudate at the infusion site and/or phlebitis of the cannulated vein. All adverse reactions which were probably or definitely related to the therapeutic regimens, including those identified by haematological and biochemical investigations, were also recorded.
Imipenem In tbe febrOe neatropenk patient
847
Table L Episodes of fever excluded from evaluation
No. of episodes treated No. (%) of protocol violations Non-haematological malignancy Incorrect dose of study drug Other antibiotics given Patient not neutropenic Patient not pyrexial Other violations No. of assessable episodes
imipenem
Treatment group combination
total
164 35(21%) 5 0 13 13 4 5
148 25 (17%) 6 2 7 8 6 1
312 60 (19%) 11 2 20 21 10 6
129
123
252
At 72 h 68 (59%) of the evaluable episodes treated with imipenem had responded compared with 65 (56%) of those treated with piperacillin and gentamicin. There were 13 and six unassessable episodes in the imipenem and combination groups, respectively. At the end-of-treatment evaluation 58 (55%) of the patients in the imipenem group were assessed as cured compared with 58 (53%) of those in the gentamicin-piperacillin group (Table IV). Four of the patient episodes in the monotherapy group and two in Table n . Demographic and clinical characteristics of the evaluable patient episodes Characteristic No. of evaluable episodes Median age (range) % of patients with neutrophil count of entry
