Severe generalized peritonitis (SGP) is frequently encountered, and the rate of mortality associated with SGP ranges from 10 to 60% (
17,
22,
24). Initial antibiotic therapy, in addition to surgery, must be started early and must cover all pathogens that can cause intra-abdominal infections (gram-negative bacilli, anaerobes, and enterococci) (
5,
17). The recent increase in the rate of multiple-antibiotic resistance among gram-negative bacilli such as
Escherichia coli justifies empiric initial treatment with a broad-spectrum antibiotic. The treatment can subsequently be adapted according to antibiotic susceptibility test results. The addition of an aminoglycoside to the treatment regimen has many theoretical advantages: (i) a broader spectrum of activity, (ii) increased synergy, (iii) increased bactericidal effect, and (iv) prevention of emergence of resistant strains (
26). However, no study that has supported the use of aminoglycosides for the treatment of SGP has been published in the literature (
14,
15). The aim of this study was to compare, by means of an equivalence trial, the effect of a broad-spectrum beta-lactam antibiotic (piperacillin-tazobactam) alone or in combination with an aminoglycoside (amikacin) in the treatment of SGP.
(This work was presented in part at the 38th Interscience Conference on Antimicrobial Agents and Chemotherapy, San Diego, Calif., 24 to 27 September 1998 [H. Dupont, C. Carbon, J. Carlet, H. Schweich, and the SGP Study, Program 38th Intersci. Conf. Antimicrob. Agents Chemother., abstr. MN-48, p. 602, 1998].)
MATERIALS AND METHODS
Participants.
This study was conducted in 35 centers in France. The trial protocol was reviewed and approved by an independent ethics committee in accordance with the ethical principles of the Declaration of Helsinki. Informed consent was obtained from each patient (or from his or her legal representative) before enrollment in the study. Eligible patients were men or nonpregnant women, ages 18 years or older, with a clinical diagnosis of complicated intra-abdominal infection according to previously published criteria (
30). Complicated intra-abdominal infection was defined by the presence of severe sepsis for patients with community-acquired infections and at least systemic inflammatory response syndrome (SIRS) for patients with postoperative or nosocomial infections. SIRS and severe sepsis were defined as reported previously (
3). SIRS was defined by two or more of the following conditions: (i) temperature, >38°C or <36°C; (ii) heart rate, >90 beats/min; (iii) respiratory rate of >20 breaths/min or partial arterial CO
2 pressure of <32 mmHg, and (iv) leukocyte count, >12,000/mm
3 or <4,000/mm
3. Severe sepsis was defined by at least one of the following criteria: (i) hypotension with systolic blood pressure of <90 mmHg; (ii) oliguria (output, ≤30 ml/h); (iii) acute alteration in mental status; (iv) coagulation abnormalities, such as a prothrombin time of <50% or a platelet count of <100,000/mm
3; (v) partial arterial O
2pressure of <60 mm Hg or partial arterial O
2pressure/fractional inhibitory O
2 pressure of <250; and (vi) blood lactate concentration, >2 mmol/liter.
Noninclusion criteria for this study included allergy to beta-lactam antibiotics or aminoglycosides, MacCabe and Jackson score of C (
19) or simplified acute physiology score (SAPS II) of >45 (
18), septic shock (
3), neutropenia (leukocyte count, <1,000/mm
3), pregnancy, nongeneralized peritonitis, and effective antimicrobial treatment given during the 30 days prior to inclusion. Patients eligible for intent-to-treat (ITT) analysis were those randomized after they gave their signed informed consent. The modified ITT (mITT) analysis population was the ITT population that presented with a surgically proven complicated intra-abdominal infection and corresponded to the main analysis population. The per-protocol (PP) population included all patients in the mITT population with no major protocol violation (exclusion criterion).
Methods.
This prospective, randomized, open-label trial was conducted to compare monotherapy (MT; piperacillin plus tazobactam) versus combined therapy (CT; piperacillin plus tazobactam and amikacin) for the treatment of SGP. Computer-generated randomization of the subjects into blocks of four subjects each was used, and each center was allocated one block of the MT treatment group and one block of the CT treatment group to ensure a stratified distribution.
No standardization of the postsurgerical approaches were planned for the protocol. Briefly, all wounds were left closed. Mesh was not usually placed on the wound, but mesh placement was left to the discretion of the surgeon. Abdomens were not irrigated in this study.
Piperacillin plus tazobactam was administered intravenously at a dose of 4 g four times daily to both groups. Amikacin was administered intravenously at a dose of 7.5 mg/kg of body weight twice daily (30-min infusion) to the CT group. The dose of amikacin could be reduced according to renal function and antibiotic concentrations in blood. The duration of treatment was at least 2 days and up to 14 days.
Successful treatment was defined by the following criteria: (i) normalization of body temperature, (ii) normalization of abdominal and systemic signs, (iii) no clinical or laboratory signs of persistent residual abscess, and (iv) no need for surgery or new antimicrobial therapy during the observation period until day 30 after the end of treatment. Clinical failure was defined by at least one of the following criteria: (i) no response or deterioration during treatment, (ii) death from any cause, (iii) any reoperation for abdominal surgery, even for wound infections, (iv) any change in antimicrobial treatment except for the use of antifungal agents, (vi) any serious adverse event related to antibiotic treatment, and (vii) the desire of the patient to stop participation in the trial.
Efficacy and failure rates were assessed at the end of treatment and after the 30-day posttreatment follow-up period by an evaluation committee blinded to the antibiotic regimen received. The primary endpoint was the failure rate for the mITT population at day 30 posttreatment. Secondary endpoints were the time to failure for the mITT population, the duration of treatment for cured patients in the mITT population, and the adverse event rate for the ITT population.
At least two sets of aerobic and anaerobic samples for blood culture were obtained by perioperative venipuncture. Peritoneal fluid samples were obtained intraoperatively. Antibiotic susceptibility was assessed by the disk diffusion method, and breakpoints were defined by the Antibiotic Susceptibility Testing Committee of the SociétéFrançaise de Microbiologie (
10).
Analysis.
Sample size estimation for the primary endpoint was based on a two-sample test of proportions with α equal to 10% and a power probability of 71%. The study was powered with a 15% tolerance around equivalence for the percent failure rate for the MT group. We calculated that at least 100 patients would be required in each treatment group to achieve this power, assuming a 35% failure rate for patients in the MT group. A total of 241 patients were recruited to compensate for possible protocol violations. The primary analysis was based on mITT analysis of all randomized patients. The failure rate was analyzed by using a logistic regression model adjusted for SAPS II (a cutoff value of 30, which corresponded to the median SAPS II for the population, was used) and type of infection (community-acquired or postoperative infection). The odds ratio for the CT group versus the MT group was estimated. The 90% confidence interval (CI) of the odds ratio was calculated and was then compared with the equivalence interval (0.54 to 1.82) calculated before the beginning of the study. Equivalence was demonstrated if the 90% CI was totally included within the equivalence interval (
2,
11).
Time to failure and duration of treatment for cured patients were analyzed by using a Cox model adjusted for both SAPS II and type of infection (community-acquired or postoperative infection). The hazard ratios for the CT group versus the MT group were estimated. The 90% CIs of the hazard ratios were calculated and then compared with the equivalence interval (0.66 to 1.53). Equivalence was demonstrated when the 90% CI was totally included within the equivalence interval.
Demographic and baseline features and adverse event rates were compared by the chi2-square test or the t test. Failure rates were analyzed by the log-rank test, and Kaplan-Meier estimates were plotted over the 30-day posttreatment observation period.
No interim analysis was planned by the protocol.
DISCUSSION
To our knowledge, this is the first study to demonstrate the strict equivalence of failure rates between treatments with a broad-spectrum beta-lactam antibiotic alone and in combination with an aminoglycoside in a population with SGP. Time to failure was also the same between the two regimens, and the duration of treatment was equivalent for cured patients. Aminoglycosides were not responsible for a higher incidence of renal failure.
The number of patients excluded after randomization following failure to give informed consent was due to the emergency procedure used to obtain informed consent with secondary assessment. However, these exclusions were well balanced between the two groups. As described above, mITT analysis was the main analysis according to the general guidelines for the evaluation of new anti-infective drugs for the treatment of intra-abdominal infections (
30). Only patients with a proven intra-abdominal infection were evaluated in this study in order to obtain a homogeneous population. The inclusion of patients with uncomplicated appendicitis (
1,
12,
24) or patients with a low Apache II score (
12,
31) is the main criticism of most studies that evaluate antibiotic regimens for peritonitis. All patients included in this study presented with SGP, with a mean SAPS II of 30, SIRS, and severe sepsis according to previously described criteria for patients with complicated intra-abdominal infections (
3). This trial was also adjusted by logistic regression for SAPS II and type of infection (community-acquired or postoperative infection).
It has been suggested that inappropriate initial antibiotic treatment for intra-abdominal infections increases the incidence of postoperative infectious events (
12,
24), the reoperation rate (
24), and the rate of mortality from community-acquired and postoperative peritonitis (
22,
24), even when antimicrobial therapy is secondarily adapted (
12,
24). It therefore appeared to be of major interest to evaluate the impact of the addition of an aminoglycoside to a broad-spectrum beta-lactam antibiotic on the overall outcomes of these severe infections.
The failure rate observed in this study (50%) was higher than that reported in previous recent studies: 30% by Cometta et al. (
9) when death was included, 18% by Barie et al. (
1), and 20% for Brismar et al. (
4). The rate of mortality from postoperative peritonitis may be as high as 50% (
22), and patients with postoperative peritonitis represented 43% of our study population. Our very strict definitions of failure may largely explain these results.
The value of an aminoglycoside in combination with a broad-spectrum beta-lactam antibiotic for intra-abdominal infections is still controversial. Broad-spectrum beta-lactam antibiotics have been successfully used alone for the treatment of intra-abdominal infections: imipenem-cilastatin (
1,
7,
9,
29), meropenem (
7), piperacillin-tazobactam (
4), and cefepime (
1). Most studies included patients with low Apache II scores and achieved treatment success rates as high as 90%; in most cases the diagnosis was appendicitis (
17,
27). Only the study by Cometta et al. (
9) compared monotherapy with combined therapy with the same beta-lactam antibiotic (imipenem-cilastatin) and netilmicin (
9). Their results, for which were for patients with peritonitis but also patients with other infections such as pneumonia and bacteremia, showed no difference in success rates between the MT group and the CT group. The addition of netilmicin increased the rate of nephrotoxicity and did not prevent the emergence of imipenem-resistant
Pseudomonas aeruginosa. No significant difference was observed in the peritonitis subgroup, but no failure due to antibiotic resistance was observed.
No studies that have compared MT versus CT during severe peritonitis are available in the literature. The equivalence design of our study allowed us to conclude that the addition of amikacin to piperacillin plus tazobactam was not justified for the treatment of SGP. In the 1980s the use of an aminoglycoside was justified by the lack of broad-spectrum beta-lactam antibiotics (
29). Recent reviews of the literature demonstrate that there is no clear evidence in support of the use of aminoglycosides for the treatment of peritonitis for the following more or less theoretical reasons (
8,
14,
17). Aminoglycosides have a good diffusion into the peritoneal cavity (
20), but they may be ineffective because their activity is reduced by the local intra-abdominal conditions of acidosis and hypoxia and the presence of drug-binding purulent debris (
21,
28). They have a marked postantibiotic effect on gram-negative bacilli, but no study has been specifically devoted to intra-abdominal infections (
16). Combination therapy with an aminoglycoside has been used to increase the spectrum of antibiotic treatment and to exert a synergistic action with beta-lactam antibiotics against most bacteria (
26). Prevention of the emergence of resistant strains of bacteria such as
P. aeruginosa has been suggested, but none of the published studies demonstrated such an effect (
9,
14,
15). Aminoglycoside treatment has also been shown to be associated with an increased incidence of nephrotoxicity (from 4 to 50%) (
9,
14,
15).
Another problem is the rates of bacteremia observed concomitantly with the peritonitis. Patients with bacteremia represent a subgroup with potentially severe illness who could obtain greater benefit from treatment with an aminoglycoside. This has been demonstrated only for neutropenic patients with bacteremia (
6). It should also be noted that in our study the percentage of patients with bacteremia was equivalent between the two regimens. Another important point in this study is the small number of patients from whom
P. aeruginosa was isolated. Isolation of
P. aeruginosadepends on the etiology of the intra-abdominal infection: for patients with community-acquired infections, it varies from 3% (
16) to 5% in our study, and for patients with postoperative infections, it varies from 11.5% in our study to 21% (
22,
25,
29). Patients with
P. aeruginosa infections might obtain greater benefit from aminoglycosides, but this has not been formally demonstrated. The emergence of enterococci may complicate treatment options during severe intra-abdominal infections. The use of an aminoglycoside, which has been validated for the treatment of enterococcal endocarditis (
13,
32), has never been validated for the treatment of clinically severe enterococcal intra-abdominal infections. Moreover, the addition of gentamicin in an experimental model of rat polymicrobial peritonitis did not increase the enterococcal killing rate (
23). However, the number of enterococci isolated in this study may be too low to demonstrate such an effect.
In conclusion, our results do not support the addition of an aminoglycoside such as amikacin to piperacillin plus tazobactam for the treatment of SGP. The value of combination therapy with aminoglycosides remains to be specifically validated for Pseudomonas spp. and enterococci during complicated intra-abdominal infections. However, in this large study that included patients with both community-acquired and postoperative or nosocomial SGP, neither of these microorganisms raised any major problems.