When the antimicrobial activity of combinations of antibiotics is studied, several results are possible:
Antimicrobial synergy, implying the beneficial interaction between two drugs exceeding simple additive effects, may take several forms (Table 9.1): one compound may potentiate the activity of the other at a biochemical level; one may assist the other to penetrate into the bacterial cell;one may protect the other from destruction; or the two compounds may act on separate sections of the bacterial population.
Table 9.1 Types of antimicrobial synergy
The clearest example of biochemical synergy is provided by the combination of trimethoprim and sulphonamides, which is most commonly used clinically as co-trimoxazole (p. 48). Sulphonamides and diaminopyrimidines such as trimeth-oprim interfere with sequential stages in bacterial folate synthesis. Sulphonamides cut off the folate supply at source, whereas trimethoprim stops regeneration of the biologically active form of the vitamin (tetrahydrofolate), which is oxidized in the course of the manufacture of thymidylic acid for nucleic acid synthesis.
Concentrations of sulphonamides and trimethoprim that are ineffective or only feebly inhibitory when used alone completely shut off of folate activity.
Another type of double blockade occurs with some β-lactam antibiotics, typified by mecillinam and cephalexin. These two agents bind to distinct proteins in the bacterial envelope (PBP 2 and PBP 3 respectively; see p. 26) and cause characteristic morphological changes in susceptible Gram-negative bacilli. Together, these compounds complement one another to elicit the spheroplast response which most other β-lactam agents evoke (see Fig. 1.5, p. 27). Most members of the β-lactam family do not exhibit this form of synergy simply because they attack both of the sites that mecillinam and cephalexin individually affect. However, the potential for this sort of interaction exists with other combinations. Imipenem and clavulanic acid, like mecillinam, bind preferentially to PBP 2 and several agents other than cephalexin (e.g. cephradine, aztreonam, and temocillin) bind almost exclusively to PBP 3. Moreover, low concentrations of most penicillins and cephalosporins have a cephalexin-like effect, so that concentrations achieved therapeutically at the site of infection could interact with agents that bind predominantly to PBP 2.
Enhancement of permeability
Many bacteria are antibiotic resistant, not because their intracellular target site is insusceptible to the drug, but because the cell manages to keep out the noxious agent. For example, the protein-synthesizing machinery of Escherichia coli is as susceptible to erythromycin as that of Staphylococcus aureus, but the drug penetrates the Esch. coli outer membrane with difficulty.
Antibiotics that act on the bacterial cell wall (e.g. β-lactam agents and glycopeptides) or the cell membrane (e.g. polymyxins) may interfere with the cell's ability to exclude another agent. The most familiar example of this is provided by the interaction of penicillins and aminoglycosides acting against enterococci. Enterococci, like other streptococci, are resistant to aminoglycosides
and the resistance is associated with failure of the drug to penetrate the cell wall. Inhibitory concentrations of penicillin kill enterococci relatively slowly, and numerous persisters survive prolonged exposure. In the additional presence of an aminoglycoside, however, the rate of killing is much increased and persisters are also killed. Studies with radioactive streptomycin have shown that this is associated with an increase in uptake of the aminoglycoside.
Aminoglycosides and β-lactam antibiotics act synergically against other species of bacteria, but the effect is not usually so dramatic as with enterococci. In the case of Gram-negative bacilli a modest enhancement of killing is usually seen. With staphylococci and streptococci the most striking effects are seen with isolates that exhibit penicillin ‘tolerance’ (p. 28).
The resistance of bacteria to β-lactam antibiotics is generally caused by bacterial enzymes (β-lactamases) that destroy the drug (p. 146). This mode of resistance suggests the possibility, mooted many years ago, of using an enzyme inhibitor to allow the enzyme-labile drug to achieve its effect unscathed. Several potent, broad-spectrum enzyme inhibitors have been developed. One of the most active is the naturally-occurring β-lactam compound clavulanic acid, which is marketed in combination with amoxycillin (co-amoxiclav) or ticarcillin.
Other useful inhibitors include the penicillin sulphones, sulbactam (available in some countries in combination with ampicillin or cefoperazone), and tazobactam (combined with piperacillin). Sulbactam is poorly absorbed when given orally, and has also been formulated as a linked ester with ampicillin to produce a so-called mutual pro-drug. The principle is similar to the one used with ampicillin pro-drugs (see p. 18).
Although resistance to aminoglycosides and chloramphenicol may be mediated by enzymic mechanisms, the protection principle has not yet been applied outside the β-lactam field.
Differential population effects
If spontaneous mutation to resistance occurs with a high frequency, two or more antibiotics may be used together mutually to prevent the emergence of resistance. If the frequency of resistance is one in every million bacteria (1 in 106) and if resistance is unlinked, the probability of double resistance occurring in a single bacterium is one in a million million (1 in 1012). If three drugs are involved, the probability becomes 1 in 1018. In order to comprehend this astronomical figure it might be useful to consider that this is equivalent to one triple mutant occurring in a tonne (1000 kg) of bacteria.
All the major antituberculosis drugs suffer from mutational resistance problems and the mutual prevention of resistance principle has been widely and successfully used for many years in the chemotherapy of tuberculosis (see Chapter 26).
Just as two antibacterial compounds can operate in a mutually beneficial way, they can also sometimes interfere with each other's activity.
Antagonism of bactericidal activity
The form of antagonism which has received most attention is that occurring between predominantly bacteristatic agents (such as tetracycline and chloram-phenicol) and those bactericidal agents (pre-eminently β-lactam antibiotics) which rely on cell growth to achieve their lethal effect. Clearly, if bacterial growth is rapidly halted, the bactericidal activity of such agents will be abolished.
In most cases antagonism operates in one direction only: one substance interferes with another, but is itself unaffected. This means that if the interfering substance is the more active of a pair, its dominance will prevail and any antagonism will be undetectable.
An exception to the general one-way rule is provided by fusidic acid and some penicillins which, when combined, exhibit mutual antagonism against certain strains of staphylococci, at least in vitro. Both fusidic acid and penicillin are usually bactericidal to Staph. aureus, but against some strains substantially less killing is observed with the combination than with either agent alone. What apparently happens is that fusidic acid prevents the growth of those cells that it does not kill and, as already stated, penicillins are unable to kill non-growing bacteria. Cell death mediated by fusidic acid is accompanied by collapse of the bacteria, suggesting a secondary effect on the bacterial cell wall following the primary inhibition of protein synthesis—perhaps because the autolytic/synthetic balance which occurs in normal wall growth is upset by the failure to produce essential enzymes. Penicillin prevents this secondary effect on the bacterial cell wall and thus interferes with the bactericidal action of fusidic acid. It is probable that a similar interaction occurs between penicillins and other inhibitors of protein synthesis in Staph. aureus.
Despite the mutual antagonism that can be demonstrated in the test-tube, there is little evidence this form of interaction has any therapeutic relevance.
In mixtures of relatively high concentrations of some β-lactam antibiotics and aminoglycosides, the β-lactam ring interacts chemically with amino groups of the aminoglycoside, so that both compounds are inactivated. The interaction probably has no significance at the concentrations achieved within the body during treatment, but these agents should not be mixed together in intravenous infusions.
Some other examples of in-vitro and in-vivo incompatibilities are given in Chapter 18.
An unusual form of antagonism occurs in erythromycin-resistant staphylococci that owe their resistance to methylation of the ribosomal binding site (p. 152). Erythromycin is a specific inducer of this enzyme, but binding of other macrolides, lincosamides, and streptogramins is prevented if the attachment site is methylated. Consequently, erythromycin can antagonize these compounds by inducing resistance to them.
The phenomenon is readily shown by disc testing of staphylococci displaying dissociated resistance. A very similar reduction of the inhibition zone produced by nalidixic acid and other quinolones can be caused by nitrofurantoin in some entero-bacteria (Fig. 9.1), but in this case the mechanism of the antagonism is unknown.
Fig. 9.1 Antibacterial antagonism and synergy. Top: one-way antagonism of nalidixic acid (right-hand disc) by nitrofurantoin (left-hand disc); bottom, synergic interaction of trimethoprim and sulphamethoxazole, shown by an enhancement of the zone of inhibition in the area between the discs, where both compounds are present. (Reproduced, with permission, from Waterworth PM, Tests of combined antibacterial action, in Laboratory Methods in Antimicrobial Chemotherapy (Reeves DS, Phillips I, Williams JD, and Wise R, eds), Churchill Livingstone, 1978).
Inducible resistance to β-lactam antibiotics, which is a feature exhibited by many strains of Enterobacter spp. and some other Gram-negative bacilli, may give rise to a similar phenomenon. The effect is associated with induction of chromosomal β-lactamases (p. 146). Potent inducers such as cefoxitin, may cause resistance to poor inducers, such as cefotaxime.
Methods for demonstrating antibiotic interactions
The most popular method used to detect antimicrobial interactions is chessboard (or checkerboard) titration, in which two drugs are cross-titrated against each other (Fig. 9.2). After incubation a so-called isobologram is constructed by plotting the inhibition of growth observed at each drug concentration on an arithmetic scale. The line of additivity joins the MICs of the individual drugs acting alone; a deviation of this line towards the axes of the graph suggests synergy (Fig. 9.3); a deviation away from the axes is often taken to indicate antagonism, although indifference may also produce this result. Alternatively, the sum of the fractional inhibitory concentrations (ΣFIC) can be calculated: if drugs A and B alone each inhibit growth at a concentration of 4 mg/l and a combination of the two inhibits growth in a mixture containing each drug at a concentration of 1 mg/l, then the ΣFIC = 1/4 + 1/4 = 1/2. A ΣFIC of 1 clearly indicates simple additivity. Theoretically, a ΣFIC below 1 should indicate synergy, but in practice partial antibacterial effects below the MIC often cause apparent deviations from additivity when true synergy is absent. It is probably wise to ignore results that indicate less than a fourfold reduction in the MIC of at least one of the components.
Fig. 9.2 Chessboard titration of two drugs. The example shows synergy: the MIC of drug A alone is 8 mg/l and that of B is 16 mg/l; in combination 2 mg A plus 4 mg B per litre inhibit growth of the test organism.
Fig. 9.3 Isobologram drawn from data presented in Fig. 9.2. The lowest concentrations of the two drugs to inhibit growth (alone and in combination) are plotted. Because the two drugs interact synergically, the resultant isobologram deviates from the line of additivity towards the axes of the graph. If the two drugs had displayed antagonism, the line would have deviated above the line of additivity. Note that, although the drugs are tested in log2 concentration steps, the results are plotted on an arithmetic scale.
Chessboard titrations detect only certain types of interaction. Interactions in which bactericidal activity is being potentiated or suppressed cannot usually be adequately investigated by this means. Although bactericidal end-points can be measured in chessboard titrations, this is an unsatisfactory indication of any interaction, since it is normally the rate of killing that is affected. Consequently. such interactions can be demonstrated effectively only by estimating the viability of the bacteria at intervals after exposure to both agents, alone and in combination, and plotting the fall in viable count over time.
Agar diffusion tests
Antimicrobial synergy and antagonism can often be demonstrated in agar diffusion tests as shown in Fig. 9.1. This type of test offers a visually satisfying method of demonstrating antibiotic interactions but is insufficiently quantitative for most purposes.
Therapeutic relevance of antibiotic interactions
Demonstration of synergy or antagonism in vitro does not guarantee that the interaction will have any relevance in treatment. Pharmacokinetic differences between agents may dictate that the compounds do not meet in the body in a ratio which has been shown to interact in the laboratory. Even when interacting compounds are carefully chosen to ensure comparable pharmacokinetics, crucial factors militating against the interaction may be overlooked. The synergy between trimetho-prim and sulphonamides is so striking when tested in chessboard titrations that the
possibility of using trimethoprim alone was scarcely considered for many years. It is now clear that, in most clinical situations, trimethoprim alone is equally effective. In urinary tract infection, a major indication for co-trimoxazole therapy, the concentration of trimethoprim achieved in urine is so far above the minimum inhibitory concentration that the slower-acting sulphonamide gets no opportunity to act. In tissue infections, in which lower concentrations may be involved, differential partition of trimethoprim and sulphonamides into intra- and extracellular compartments may prevent their interaction.
In fact, only a few antimicrobial interactions have been proven to have therapeutic relevance. These include: the interaction between penicillin and aminogly-cosides in enterococcal endocarditis; the avoidance of resistance by triple therapy in tuberculosis; the protection of β-lactam agents by β-lactamase inhibitors; and the antagonism between tetracycline and penicillin in pneumococcal meningitis.