Rationale for Use of Antimicrobial Combinations in
Treatment of Gram-Negative Infections
A Review of Recent Reviews
JAMES J. RAHAL, Jr., M.D.
New York, New York
From the Infectious Disease Section, New York
Veterans Administration Medical Center and the
Department of Medicine, New York University
Medical Center, New York, New York. Requests
for reprints should be addressed to Dr. James J.
Rahal. Jr., Veterans Administration Medical
Center, 408 First Avenue, New York, New York
10010.
68 August 29, 1983 The American Journal of Medicine
All investigators apparehtly agree that the most common and
compelling reason for using more than one antibiotic to treat a single
organism is to achieve a bactericidal effect. Most studies, both re-
trospective and prospective, have demonstrated that two effective
antibiotics yield better results than one in neutropenic patients
and/or those with rapidly fatal underlying disease, despite the ab-
sence of consistent in vitro synergy. Bacteremias caused by Pseu-
domona aeruginosa or Klebsiella pneumoniae may be benefited most
by synergistic combinations. This may not be true for patients with
non-neoplastic disease and normal granulocyte counts, or for pa-
tients infected with other species of gram-negative bacilli. Syner-
gistic bactericidal activity is necessary for the successful therapy
of endocarditis due to P. aeruginosa, but it may not assure success.
The systemic immunodefkiency of neutropenic patients may parallel
a localized immunodeficiency in endocarditis, since leukocytes are
not effectively mobilized to the site of infection in endocarditis.
Antagonistic antibiotic combinations are likely to be particularly
harmful in neutropenic patients.
Since my first personal encounter with the clinical relevance of anti-
biotic synergy and antagonism in 1976 [ 11, several comprehensive
and critical reviews have added further perspective to the rational use
of antibiotic combinations [2-41. My purpose will be to provide the
reader with a review of those reviews in order to collate the therapeutic
principles that have emerged regarding the use of antibiotic combi-
nations in the treatment of gram-negative infections.
All investigators appear to agree that the most common and com-
pelling reason to use more than one antibiotic to treat a single or-
ganism is to achieve a synergistic bactericidal effect. The use of
combinations as initial therapy for unknown infection, for known mixed
infections, and for the prevention of resistance are subjects for sep-
arate discussions and will not be dealt with here.
ANTIBIOTIC SYNERGY
Laboratory methods most commonly used to define antibiotic synergy
are the checkerboard tube dilution and killing rate techniques. In the
former, synergy is defined as the achievement of bactericidal activity
by two agents at no more than one-fourth of the minimal bactericidal
concentration of each [ 21. Synergy by the latter technique has been
defined as a hundredfold decrease in viable organisms after 24 hours
 AMDINOCILLIN SYMPOSIUM-RAHAL

of incubation in the presence of two agents as com- effective, whereas treatment with antipseudomonal
pared with either one alone at the same concentration penicillins alone against P. aeruginosa has been less
[ 21. Norden [ 5] has demonstrated poor correlation effective. Therapy with cephalosporins or aminogly-
between these techniques when the same organisms cosides alone has been least effective. Some investi-
are tested, and Moellering [6] has reviewed the pitfalls gators have suggested that poor results with amino-
in the interpretation of such in vitro data. Others, how- glycosides alone may be related to inadequate serum
ever, have noted that it is very unusual for one method concentrations [ 31.
to indicate synergy and another antagonism when the In non-neutropenic cancer patients, Klastersky et al
same antibiotics and organisms are used [3]. Further- [8] demonstrated that the combination of carbenicillin
more, the majority of studies (both retrospective and plus gentamicin was significantly more effective than
prospective) designed to compare the efficacy of dif- gentamicin alone against all gram-negative infections.
ferent antibiotic regimens in the therapy of gram-neg- Carbenicillin alone was less effective than the combi-
ative bacteremia in neutropenic patients have demon- nation, but without statistical significance. Klastersky
strated improved results when two effective agents are and colleagues [9] next compared regimens of peni-
used rather than one, despite the absence of consistent cillin or carbenicillin, combined with amikacin, in 117
in vitro synergy [2,3]. Other studies have correlated an patients with serious neoplastic disease who had un-
improved outcome following combination therapy that dergone radiation therapy or who had received cytotoxic
has yielded bactericidal activity in 1:8 or 1:16 dilutions or corticosteroid therapy but who were not neutropenic
of the patient’s serum. In several such studies, in vitro (granulocyte count greater than 1,000/mm3). Among
synergy has not been documented. These results all patients, and particularly among those who were
suggest the possibility that true in vitro synergy (as bacteremic, treatment with a combination exhibiting in
defined herein) may not be a requisite for the achieve- vitro synergy against the infecting pathogen yielded a
ment of beneficial results by antibiotic combinations in significantly greater response rate. Also, a favorable
neutropenic patients. Young [2] has defined synergy outcome was associated with bactericidal serum ac-
as that condition that exists when the effect of two tivity of greater than or equal to 1:8. Among narcotic
agents in combination is greater than the sum of their addicts with P. aeruginosa endocarditis, Reyes et al [lo]
individual activities. Klastersky and Zinner [3] state that found that synergistic bactericidal activity was neces-
synergy is an antibacterial effect beyond that which can sary for successful therapy, but did not assure suc-
be predicted on the basis of their individual activities. cess.
Strict adherence to synergy as a criterion for a benefi- In summary, in the therapy of bacteremia in neutro-
cial effect would exclude the possibility that an additive penic patients, and those with serious neoplastic dis-
effect may yield clinical benefit when a maximum dose ease, antibiotic combinations are beneficial, with or
of each drug is used. In such instances, toxicity may without demonstrable synergy. P. aeruginosa and K.
preclude the achievement of the same benefit from a pneumoniae bacteremias may be benefited most by
single agent given in a larger dose. It may thus be rea- synergistic combinations. No strong evidence exists to
sonable to study possible correlations between clinical indicate that this is true for patients with non-neoplastic
results and additive, as well as synergistic in vitro ac- disease and normal granulocyte counts, or for patients
tivity [3 ] In doing so, an additive effect must be dis- infected with other species of gram-negative bacilli. P.
tinguished from an indifferent result in which the com- aeruginosa endocarditis in heroin addicts appears to be
bined activity of two drugs is no greater than that of ei- an exception.
ther one alone at the same concentration.
SINGLE VERSUS COMBINED ANTIBIOTIC THERAPY

COMBINED ANTIBIOTIC THERAPY Since leukocytes are not effectively mobilized to the
site of infection in endocarditis, an analogy may be
As stated, clinical studies have consistently demon- drawn between the systemic immunodeficiency of
strated that two effective antibiotics yield better results neutropenic patients and a localized immunodeficiency
than one in neutropenic patients and/or those with in endocarditis, both of which require bactericidal an-
rapidly fatal underlying disease. Young [2] has em- timicrobial therapy and benefit from synergistic (and
phasized that this is a distinct reversal of earlier findings possibly additive) combinations [ 11 1. Recent studies
by McCabe and Jackson [ 71 who found that antibiotic on host defense mechanisms in bacterial meningitis
therapy did little to alter outcome in such persons. Most suggest that a similar localized immunodeficiency due
data in this area involve the treatment of P. aeruginosa to defective opsonization of encapsulated organisms
or K. pneumoniae bacteremia with an antipseudomonal exists in the subarachnoid space [ 121. Therapy of
penicillin or cephalosporin plus an aminoglycoside. For gram-negative bacillary meningitis with either single or
these infections, combination therapy has been most combined antibiotics has yielded a bacteriologic re-

August 29, 1983 The American Journal of Medicine 69

AMDINOCILLIN SYNPOSIUK-RAHA:

sponse in 60 to 65 percent of the patients at best [ 121. tremely important determinant of outcome. Many
Recent studies suggest that single drug therapy with studies carried out in neutropenic patients fail to de-
moxalactam for meningitis due to susceptible enteric scribe the proportion of patients in whom leukocyte
bacilli has improved the outcome of this infection [ 131. counts return to normal during therapy. Another factor
Whether combination therapy may improve upon the not frequently considered in evaluating the results of
results obtained with highly potent new agents alone clinical trials is the sample size, particularly in studies
remains to be determined [ 141. One may further that demonstrate a trend without statistical significance.
speculate that other localized infections may benefit In such instances, larger studies may be required to
from combination therapy that provides enhanced reach significance. Unless the size of the benefit sought
bactericidal activity. Such infections may be charac- and the number of patients needed to demonstrate it are
terized by local impairment of host responses due to planned, studies that fail to prove a significant difference
infarction or other defects in circulation. Gram-negative may also fail to prove the absence of such a difference.
bacillary pneumonia and osteomyelitis are notorious for Although this approach may require very large study
their poor response to any form of treatment, and populations, Sacks et al [ 181 have suggested an al-
combination therapy is widely recommended for both ternative. They propose reconsideration of a p value
[ 15,161. However, its benefit as compared with sin- less than 0.05 as the point at which a difference is
gle-drug treatment has not been documented by com- considered significant. If small differences are suffi-
parative studies. A corollary to the beneficial effects of ciently important clinically, then their demonstration with
additive or synergistic antibiotic combinations in neu- 85 to 90 percent confidence may be of value.
tropenic patients may be that antagonistic combinations
are likely to be harmful in such patients [ 11. This may CONCLUSION
also apply to other situations in which local deficiencies
of host response exist, for example, endocarditis, Recently published reviews have supported my initial
meningitis, necrotizing pneumonia, and osteomye- optimistic view of the clinical relevance of antibiotic
litis. synergy [ 11. Because maximum doses of antibiotics are
Young [2] has emphasized those variables that must usually employed for critically ill patients, additive ef-
be considered when single and combined antibiotic fects of drug combinations are likely to prove beneficial
therapies are compared. They may be categorized as as well. These benefits are most clearly evident in
follows: neutropenic patients, but they may also exist in certain
(A) Patient selection: (1) underlying disease: non- localized infections with suboptimal host defenses.
fatal, ultimately fatal, rapidly fatal; (2) presence and Newer, more potent bactericidal agents, when admin-
duration of neutropenia; (3) presence of bacteremia; (4) istered alone, may be as effective as combination
tissue source of infection; (5) pathogenic organism. therapy. However, the diminished bactericidal activity
(B) Study logistics: (1) dose and route of antibiotic: of almost all beta-lactams against large inocula of P.
(2) blood or other body fluid concentrations achieved; aeruginosa warrants caution with this approach. I agree
(3) incidence of resistance to one or more study drugs; with Young’s suggestion that single-drug trials with new
(4) use of leukocyte transfusions. agents be carried out first in immunocompetent hosts
The findings of Love et al [ 171 in a four-year study [2]. In addition, these newer agents warrant testing in
of bacteremia in neutropenic patients emphasize that combination with other drugs in the treatment of neu-
recovery from neutropenia during therapy is an ex- tropenic patients [ 191.

REFERENCES
1. Rahal JJ Jr: Antibiotic combinations: the clinical relevance
of synergy and antagonism. Medicine (Baltimore) 1976; 57:
179-195.
2. Young LS: Combination or single drug therapy for gram-
negative sepsis. Curr Clin Top Infect Dis 1982; 3: 177-
204.
3. Klastersky J, Zinner SH: Synergistic combinations of antibi-
otics in gram-negative bacillary infections. Rev Infect Dis
1982; 4: 294-301.
4. Eliopoulos GM, Moellering RC Jr: Antibiotic synergism and
antimicrobial combinations in clinical infections, Rev Infect
Dis 1982; 4: 282-293.
5. Norden CW: Problems in determination of antibiotic synergism
in vitro. Rev Infect Dis 1982; 4: 276-280.
6. Moellering RC Jr: Antimicrobial synergism-an elusive con-
cept. J Infect Dis 1979; 140: 639-641.
7. McCabe WR, Jackson GG: Gram-negative bacteremia Il.
Clinical, laboratory and therapeutic observations. Arch
Intern Med 1962; 121: 856-864.
8. Klastersky J, Cappel R, Daneau D: Therapy with carbenicillin
and gentamicin for patients with cancer and severe infec-
tions caused by gram-negative rods. Cancer 1973; 31:
33 l-336.
9. Klastersky J, Meunier-Carpentier F, Prevost JM: Significance

70 August 29, 1983 The American Journal of Medicine

 of antimicrobial synergism for the outcome of gram-neg-
ative sepsis. Am J Med Sci 1977; 273: 157-167. 15.
10. Reyes MP, El-Khatib MR, Brown WJ, Smith F, Lerner AM:
Synergy between carbenicillin and aminoglycoside (gen-
tamicin or tobramycin) against Pseudomonas aeruginosa 16.
isolated from patients with endocarditis and sensitivity of
isolates to normal human serum. J Infect Dis 1979; 140:
192-202.
11. Ernst JD, Sande MA: Antibiotic combinations in experimental 17.
infections in animals. Rev Infect Dis 1982; 4: 302-310.
12. Rahal JJ Jr, Simberkoff MS: Host defense and antimicrobial
therapy in aduft gram-negative bacillary meningitis. Ann 18.
Intern Med 1982; 96: 468-474.
13. Rahal JJ Jr: Moxalactam therapy for gram-negative bacillary
meningitis. Rev Infect Dis 1982; 4: S606-S609. 19.
14. Ccxmdo ML, Gombert K, Charebin CE: Designing appropriate
therapy in the treatment of gram-negative bacillary men-
August 29, 1983
ingitis. JAMA 1982; 248: 71-74.
Sande MA: Antimicrobial therapy for two serious bacterial
infections. Enterococcal endocarditis and nosocomial
pneumonia. Arch Intern Mad 1982: 142: 2033-2044.
Norden CN, Shaffer MA: Activities of tobramycin and azfocillin
alone and in combination against experimental os-
teomyelitis caused by Pseudomonas aeruginosa. Antimi-
crab Agents Chemother 1982; 21: 62-65.
Love W, Schimpff SC, Schiffer CA, Wiernik PH: Improved
prognosis for ganulocytcpenic patients with gram-negative
bacteremia. Am J Med 1980; 68: 643-648.
Sacks H, Chalmers TC. Smith l-l Jr: Randomized versus his-
torical controls for clinical trials. Am J Med 1982; 72:
233-240.
Klastersky J: Treatment of severe infections in patients with
cancer. The role of new acyl-penicillins. Arch Intern Med
1982; 142: 1984-1987.
The American Journal of Medlclne 71