Antimicrobial Chemotherapy, 4th Edition

Laboratory aspects of antimicrobial therapy


Antibiotic sensitivity testing and assay

  1. Greenwood
  2. C. B. Slack

Sensitivity testing

Purpose of sensitivity testing

Since therapy of infection normally begins, quite properly, before laboratory results are available, antibiotic sensitivity testing primarily plays a supplementary role in confirming that the organism is susceptible to the agent that is being used. Sometimes it may enable the clinician to change from a toxic to a less toxic agent, or from an expensive to a cheaper one.

Usually the laboratory report will influence treatment only if the patient is failing to respond. By this time, the laboratory should have succeeded in establishing the sensitivity pattern of the offending organism (if it is bacterial) sufficiently for the clinician to be able to make an informed decision as to how treatment might be modified. Sensitivity testing of non-bacterial pathogens is not usually possible, although limited antifungal testing is carried out in some centres, and methods for testing antiviral agents are being developed.

The laboratory also has an important public health function in recording and disseminating data on the sensitivity patterns of common pathogens in the hospital and in the community, in order that reliable predictions of their probable sensitivity may be made. Patterns of bacterial sensitivity and resistance vary considerably from place to place and hospitals, or even wards, often have their own particular resistance problems. It is, therefore, important that each hospital keeps its own record of resistance trends. Global patterns and trends can be monitored by laboratory networks reporting to a central point.

Finally, sensitivity testing is used to establish the degree and spectrum of in-vitro activity of new antibacterial agents first of all in the laboratories of the pharmaceutical houses from whence most new developments emanate, but also


in diagnostic laboratories where the new agent can be tested against the various types of organism encountered locally.

Methods of testing

The antibiotic sensitivity of bacteria can be assessed in a variety of ways according to individual preference, the constraints of cost, the nature of the bacterium, the number of strains requiring investigation and the degree of accuracy required. Traditional methods fall into one of three main categories:

  • Agar diffusion tests, in which the antibiotic is allowed to diffuse from a point source, commonly in the form of an impregnated filter paper disc, into an agar medium that has been seeded with the test organism.
  • Broth dilution tests, in which serial (usually twofold) dilutions of antibiotic in a suitable fluid medium are inoculated with the test organism. The highest dilution of the antibiotic to prevent the development of visible growth after overnight incubation is the minimum inhibitory concentration (MIC).
  • Agar incorporation tests, which are essentially similar to broth dilution tests except that the antibiotic dilutions are incorporated in an agar medium in a series of Petri dishes. These are spot-inoculated with a number of test organisms, usually by means of a semi-automatic inoculating device.

Agar diffusion tests

Most diagnostic microbiology laboratories test antibiotic sensitivity of bacteria by some form of agar diffusion test in which the organism under investigation is exposed to a diffusion gradient of antibiotic provided by an impregnated disc of filter paper. When the bacterial population reaches a certain critical concentration, no further inhibition of growth can be achieved and the edge of an inhibition zone is formed. Up to six antibiotics can be tested on one culture plate.

Several versions of the disc diffusion test are in use in different countries, but none has achieved universal approval. Various standardized protocols have been described, and these, if properly followed, allow a highly reproducible determination of sensitivity.

The formation of inhibition zones represents the dynamic interaction between antibiotic diffusion and bacterial growth. Within certain limits the size of the inhibition zone is a measure of the MIC of the antibiotic for the test organism, but for individual strains the relationship between the two may be far from perfect. By examining numerous strains of known MIC, regression analysis can be applied and the relationship quantified for each bacterial species and anti- biotic (Fig. 8.1).


Fig. 8.1 Hypothetical example of correlation of disc diffusion zone sizes and minimum inhibitory concentration (MIC) values. Note that MIC values are plotted on a logarithmic scale and appear in discrete log2 steps, whereas disc zone sizes are measured on a continuous arithmetic scale. A zone size of 6 mm indicates no zone of inhibition, since this is the diameter of the antibiotic-containing disc. The example shown indicates that most Staph. aureus strains are fully susceptible to the antibiotic, but a few strains are fully resistant; allPs. aeruginosa strains are resistant. A very wide spread of susceptibility values is observed with enterobacteria, and correlation is less than perfect. It is this group that presents interpretative difficulties.

A specialized version of the agar diffusion test that has been introduced commercially is the ‘Etest’. A plastic strip with a linearly decreasing concen- tration of antibiotic is applied to the surface of an inoculated agar plate. An elliptical zone of inhibition is formed after incubation, and the point at which


this intersects with the strip corresponds to the MIC of the antibiotic, which can then be read off from graduations on the strip. The test is suitable for many types of organisms and provides a simple, if expensive, means of assessing the MIC if this is required.

Factors affecting disc tests

Many of the factors determining the outcome of disc tests are the same as those encountered in other methods of sensitivity testing (see below: Factors affecting sensitivity tests), but some are peculiar to this method.

The size of the inhibition zone may be profoundly influenced by the solubility, ionic charge, and molecular size of the antibacterial agent. Large cyclic peptides, such as the polymyxins, diffuse poorly and produce small zones of inhibition even when sensitive organisms are tested against high drug concentrations.

The growth rate of the bacterial cell will also affect zone sizes, slow-growing organisms giving rise to large zones. A corollary of this is that if cultures are


allowed to stand at room temperature before incubation at 37 °C an overestimate of bacterial susceptibility may be made.

One of the most difficult factors to control adequately is the initial amount of antibiotic in the disc. Commercially available discs usually contain the stated amount of drug with a tolerance of 67–150 per cent; i.e. a disc rated at 30 µg may contain 20–45 µg of drug. If discs are incorrectly stored, allowed to become moist, or used beyond the expiry date, they may contain much less than the stated value. Note that disc content is described in µg per disc, which should not be confused with µg per ml, often used to specify antibacterial potency. In fact, in disc testing the concentration of antibiotic in µg per ml in the area immediately surrounding the disc can be a much higher value than that of the nominal disc content.

Despite these and other variables, the disc diffusion test is widely used for antibiotic sensitivity tests for reasons of speed, flexibility, simplicity, and cost. Under carefully controlled conditions it is capable of producing satisfactory results with the more common rapidly growing pathogens such as Staphylococcus aureus, Pseudomonas aeruginosa, and various enterobacteria; it is less suitable for fastidious or slow-growing bacteria such as anaerobes, streptococci, Haemophilus spp., and Neisseria spp., for which alternative procedures are preferable.

Control of disc diffusion tests

Since zone sizes are affected by many variable factors it is essential that all tests include adequate controls. A common method of control is to include a series of plates inoculated with a standardized inoculum of organisms of known sensi- tivity along with each batch of tests and to measure the size of the inhibition zones produced. This adequately controls culture medium composition (provided the same batch of medium is used for all tests) and incubation conditions, but makes no allowance for variations in disc content.

Disc content is more satisfactorily controlled by Stokes’ method, in which both test and control organisms are inoculated on the same culture plate so that a direct comparison of zone size and inoculum density can be made (Fig. 8.2). With this method control and test strains should be of the same species and have similar growth characteristics on the medium used. Critics of the method point to the difficulty of reliably achieving these conditions in routine practice.


Fig. 8.2 Disc sensitivity testing by Stokes' method, in which inhibition of the test organism by antibiotics is directly compared with that of a similar control organism of known sensitivity. In the example shown, the test organism (Staph. aureus) is resistant to penicillin (right-hand disc), but fully sensitive to erythromycin (left-hand disc). (Photograph courtesy of George Sharp and Richard Edwards, Nottingham Public Health Laboratory.)

Stokes’ method is particularly valuable in circumstances in which disc potency is uncertain; e.g. in countries in which supply is unreliable, refrigeration inadequate, and the climate humid.

Broth dilution tests

Conventional test

Conventional broth dilution tests are expensive in time and materials and tend to be used when only a few strains of bacteria need to be tested or when an accurate MIC estimation is required. A series of twofold dilutions of the antibiotic under


study is prepared in a 1 ml volume of a suitable broth medium and a standard inoculum of the test strain (commonly 105 bacteria) is introduced into each tube. In a variant microdilution method 0.1 ml volumes in microtitration trays (often preprepared with freeze-dried antibiotic) are used. The test is incubated at 37 °C overnight and the end-point is read as that concentration of antibiotic in which no turbidity can be seen. Uninoculated tubes containing broth plus antibiotic and broth alone act as sterility controls; an antibiotic-free tube inoculated with the test organism serves to indicate that the organism is viable in case the end-point is missed. The whole test is controlled by a parallel titration of the antibiotic against an organism of known sensitivity (Fig. 8.3).


Fig. 8.3 Broth dilution test of antibiotic susceptibility. Twofold dilutions of antibiotic are inoculated with a known number of organisms and incubated at 37 °C overnight. The potency of the antibiotic is checked by titrating it against a control organism of known susceptibility. The minimum inhibitory concentration (MIC) is the highest dilution in which no turbidity develops (2 mg/l for the control organism and 8 mg/l for the test organism in the example shown).



Determination of MBC

A major advantage claimed for broth dilution procedures is that the minimum bactericidal concentration (MBC) of antibiotic for the test strain may additionally be determined if so desired. This is done by subculturing on to solid medium a standard volume of broth from those antibiotic dilutions showing no visible growth after overnight incubation. The number of colonies developing after a further overnight period of incubation is compared with the number originally inoculated. The MBC is arbitrarily defined as the lowest concentration that kills 99.9 per cent of the original inoculum (reducing bacterial numbers 1000-fold); e.g., if 105 bacteria were inoculated into 1 ml of broth, a 0.1 ml volume subcultured on to solid medium should yield less than 10 colonies if the antibacterial agent is bactericidal.

Rate of kill

The MBC is a very crude estimate of bactericidal potency which reveals nothing about the kinetics of bacterial response. It is more satisfactory, but much more laborious, to perform sequential viable counts at regular intervals after exposure to various concentrations of the drug. This indicates the rate of bacterial killing and may detect recovery during the overnight incubation period which would be masked in the overnight end-point. It should be noted that viable counting methods detect colony-forming units, not absolute numbers of bacteria since chains and clumps give rise to single colonies.

Agar incorporation tests

In this method the antibiotic dilutions are made in solid agar medium by adding antibiotics to molten agar (at 45 °C) and pouring into Petri dishes. The test strains are inoculated on to the surface of the medium. This is the preferred method when large numbers of strains are to be examined since 20 or more strains can be accommodated on a single plate. The inoculum is applied with a multipoint inoculator which transfers a small drop of broth culture to each of the desired number of antibiotic-containing plates. A suitable inoculum yields a barely confluent spot of culture after overnight incubation on an antibiotic-free control plate. A series of control strains of known sensitivity are included on each plate.



Break-point tests

By using an appropriate series of antibiotic concentrations, MICs can be estimated by the agar incorporation method. The results are commonly found to be one or two twofold dilutions lower than those obtained by broth dilution. Since accurate MICs are seldom required for routine sensitivity tests the range of dilutions can be considerably reduced by using only one or two preselected break-point concentrations related to agreed cut-off points of sensitivity or resis- tance (Table 8.1). Like the content of antibiotic discs these break-points are somewhat arbitrary, but are related to levels achievable in serum, tissue, or urine during therapy.

Table 8.1 Example of antibiotic susceptibility testing by the break-point method



Ampicillin concentration


Test organism

Plate 1 (nil)

Plate 2 (8 mg/l)

Plate 3 (64 mg/l)





No growth

No growth

Fully sensitive




No growth

Reduced sensitivitya





Fully resistant


No growth

No growth

No growth

Void testb


Susceptible to levels achievable in urine.

Control failed to grow.


Mycobacterium tuberculosis is slow growing and needs special media for cultivation. Traditionally, sensitivity testing has been carried out by the resistance ratio method, a variety of the agar incorporation test in which growth of the organisms on medium containing antimycobacterial agents is compared with that of a control. The method is cumbersome and an alternative technique, in which inhibition of growth is detected by failure to release radioactive carbon dioxide from a labelled substrate, is gaining popularity. This method can be automated and produces reliable results within 7 days. Molecular techniques aimed at the rapid detection of resistance genes are under development.

  1. lepraecan be grown only in certain experimental animals and sensitivity tests are not routinely available. So-called ‘atypical’ (environmental) mycobacteria encompass a heterogeneous collection of organisms with different growth characteristics, some of which can be tested by methods similar to those used for M. tuberculosis.



Other problems

As well as the problems provided by slow-growing and nutritionally fastidious organisms, some resistance mechanisms present particular difficulties, and special methods may be needed if they are to be detected reliably. These include the detection of resistance to β-lactam agents in staphylococci and pneumococci associated with alterations in penicillin-binding proteins (see p. 152); high-level aminoglycoside resistance and vancomycin resistance in enterococci; and reduced susceptibility to glycopeptides in staphylococci.

Factors affecting sensitivity tests

Inoculum size

The most important single factor affecting the result of sensitivity tests is the number of bacteria present in the original inoculum; this must be carefully standardized if reliable and reproducible results are to be obtained. Unfortunately, the therapeutically ‘correct’ inoculum has never been precisely defined and the inocula usually used are based on tradition rather than firm evidence.

Composition of culture medium

Culture media used for sensitivity testing must be free of antibiotic antagonists and readily support growth of the bacteria under test. Several formulations of medium suitable for sensitivity testing of a wide range of bacterial pathogens are commercially available. Mueller–Hinton medium has become established for the purpose in the US, but it is unclear that this has any special properties which single it out as being particularly suitable, and other specially formulated sensitivity test media are usually preferred elsewhere. Media that contain thymidine are unsuitable for testing trimethoprim or sulphonamides, but the addition of lysed horse blood to the medium renders it suitable because horse blood contains the enzyme thymidine phosphorylase.

As well as being able to support growth of the major pathogens well, sensitivity test media should be standardized for carbohydrate content, pH, osmolality, and cation content. Excess fermentable carbohydrate will cause a change in pH of the medium during growth, which may alter the rate of growth of the bacteria and the activity of the antibiotic. The aminoglycosides and erythromycin, for example, are much less active in an acid than in an alkaline medium, whilst nitrofurantoin is more active at an acid pH. Aminoglycosides may be affected by a number of different constituents of culture media including divalent cations (calcium and magnesium), sodium chloride, and phosphates. These factors may differentially affect the activity against different species of bacteria so that, for example, the activity of aminoglycosides against Ps. aeruginosa is particularly affected by alterations in divalent cations.




The bactericidal activity of β-lactam antibiotics, which in general rely on osmotic rupture of the bacterial cell to achieve their lethal effect, is markedly influenced by the osmolality of the growth medium. Species of bacteria which have a naturally low internal osmolality, such asProteus mirabilis and Haemophilus influenzae, are predominantly affected bacteristatically by β-lactam agents unless the osmolality of broth media is artificially reduced to below physiological levels.

Antibiotic formulation

An obvious factor which is sometimes overlooked in sensitivity testing is that the formulation of the antibiotic used must be appropriate. Esters of ampicillin, carbenicillin, mecillinam, cefuroxime, and erythromycin are antibacterially inactive pro-drugs that release the active parent compound in the body. Similarly, chlo-ramphenicol succinate and clindamycin phosphate are inactive in vitro and the parent compounds should be used in laboratory tests. Sulphomethyl derivatives of polymyxins are much less active in vitro than are the non-sulphomethylated varieties, although they spontaneously break down to the more active parent form on incubation.

Newer methods of sensitivity testing

Alternative methods of assessing bacterial susceptibility to antibiotics are gradually coming into use in an effort to obtain results more quickly than by traditional procedures. Most rapid of all are the tests which detect resistance to β-lactam antibiotics by the demonstration of β-lactamase activity. Several methods are available which will accomplish this in a few minutes once a bacterial culture is available. However, since β-lactam agents display differential susceptibility to the various β-lactamases, the tests are of limited value. They have been successfully used to detect β-lactamase-mediated resistance to penicillin and ampicillin in Neisseria gonorrhoeae and H. influenzae, which is due to one particular variety of β-lactamase (TEM-1, see p. 148).

Of wider applicability are those techniques which employ turbidimetry or fluo-rimetry to detect antibacterial activity by comparing the growth of bacteria exposed to antibiotic with a drug-free control over a time span which for fast-growing organisms can be as little as 2–3 h. There are several commercial devices that do this with various degrees of sophistication, and these semi-automated methods are growing in popularity in some countries. The results show discrepancies with more traditional methods for certain bacterium/drug combinations and there is controversy about the correct interpretation of these.

An alternative approach that is gaining in popularity is the use of molecular methods that are able to detect DNA sequences associated with resistance traits. These provide a specific and reliable means of detecting antibiotic resistance, although presence of the gene does not necessarily mean that it will be expressed.


Because of the large array of resistance mechanisms that may be encountered, and other technical problems, it is likely to be some time before these methods can be economically introduced into routine laboratory practice. They may have a particular value with problem organisms or for the detection of specific resistance mechanisms.

Clinical relevance of antibiotic sensitivity tests

A laboratory report of sensitivity or resistance by no means guarantees that the results will translate into clinical success or failure if the agent is used in therapy. Patients may fail to respond to antibiotics judged to be fully active against the offending microbe, or may recover despite the use of agents to which the organism is resistant. These situations arise because laboratory tests offer relatively crude estimates of susceptibility that fail to take into account many crucial features of the infection in the patient (Table 8.2).

Table 8.2 Some aspects of infection in vivo that may cause the results of in-vitro tests not to be reflected during treatment



Phenotype different in vivo

Growth rate different in vivo

May be adherent (e.g. device-related)

Location of infection

Intracellular versus extracellular

Body compartment (CSF, lung, urine, etc.)

May be associated with large collection of pus

May be associated with foreign body

May be within biofilm

Host immune response (and other defence mechanisms)

Contributes to recovery

May be influenced by disease

May be influenced by drugs (including antibiotics)

Pharmacology of drug

Pharmacokinetic behaviour

Distribution (intracellular or extracellular)

Metabolism in vivo

Penetration to the site of infection (CSF, pus, sputum etc.)


CSF, cerebrospinal fluid

None the less, antibiotic sensitivity testing offers a generally reliable guide to therapy, particularly in the seriously ill patient in whom laboratory tests may provide an indispensable guide to patient care.

Antibiotic assay

In the context of antimicrobial chemotherapy, antibiotic assay means the estimation of antibiotic concentrations in serum, urine, or other body fluids at


appropriate times after giving the drug. The indications to assay antimicrobial drugs in biological material in routine practice are few. However, in the development of a new drug, determination of the pharmacokinetic profile in health and disease forms an important part of the evaluation of the agent, and frequent assays in various body fluids are part of this process.

In the management of individual treatment, determination of antibiotic concentration in blood plasma is usually only necessary on two counts: first, with drugs of known toxicity where the adverse effect is dose related; second, to monitor efficacy when there is a narrow therapeutic range between adequate and toxic levels (Fig. 8.4). Occasionally, it may be useful to assay antibiotic levels in other fluids when there is reason to doubt that treatment is achieving adequate levels, for example in cerebrospinal fluid (CSF), but this is seldom done.


Fig. 8.4 The concept of therapeutic range. An antibiotic administered to a patient at time 0 reaches a peak concentration in plasma and is subsequently eliminated. The therapeutic concentration needed to achieve an antibacterial effect (usually taken as the MIC for the infecting organism) is shown by the lower dashed line; the concentration above which toxic side-effects are known to be commonly encountered is shown by the upper dashed line. The difference between these values is the therapeutic range. The therapeutic index is the ratio of the toxic concentration to the therapeutic concentration.

In the management of some infections, of which tuberculosis is the most important, it may be necessary to check for compliance. In such cases, an indication of whether the patient is taking the medication is all that is required and the assay needs to detect only the presence of the drug rather than an accurate concentration. It is good practice to monitor compliance during the months of therapy with antituberculosis drugs; rifampicin can be easily detected in urine by colorimetric methods and several qualitative tests are available for the detection of isoniazid or its metabolites.



A special form of assay is sometimes used in the treatment of certain infections such as bacterial endocarditis, in which it is important to achieve bactericidal levels of drug. In this case a sample of the patient's serum, obtained at a period of time after administration of the antibiotic at which a peak concentration is anticipated, is titrated against the organism responsible for the infection (so-called ‘back-titration’). Since the object is to establish that a sufficiently high bactericidal titre is maintained it is necessary to measure the bactericidal, not just the bacteristatic end-point.

Assay of aminoglycosides

In practice, aminoglycosides, and gentamicin in particular, are the most commonly assayed antibiotics in hospital laboratories, because of the unpleasant nephro-toxic and ototoxic side-effects associated with these drugs (see Chapter 18). Assays should be performed if treatment is for longer than 48 h, particularly if there is any renal impairment, and always in older patients. Indeed, it might be considered negligent if a patient developed side-effects attributable to amino-glycosides and the drug concentrations had not been monitored.

The therapeutic range of gentamicin concentrations in serum was originally thought to be 2–10 mg/l, but it is far from certain that single high peak concentrations correlate simply with adverse effects. What may be of more importance is the area under the curve, i.e. the total concentration of drug related to time. Thus, relatively minor increases in pre-dose ‘trough’ levels, such as occur with minimal renal impairment, may be of more importance than high peaks.

Standard dosage regimens

Standard dosage regimens of aminoglycosides for the acutely ill patient are often established by use of a sliding-scale guide (nomogram) which requires information on the age, sex and body weight of the patient as well as serum creatinine as a measure of renal function. Even with accurate use, nomograms tend to underdose for children, people with cystic fibrosis and some renal patients, and to overdose the elderly. Regular monitoring by serum assays are, therefore, still required. If doses are to be given every 8 or 12 h, a pre-dose (trough) concentration should be <2 mg/l and a 1 h post-dose ‘peak’ in the range 5–10 mg/l. The situation should be monitored after about the third dose and twice a week thereafter—or more frequently if there is renal impairment or instability of concentrations.

Single daily dosage regimens

It is now more common to administer aminoglycosides in higher but less frequent doses; the whole of the standard daily dose is given in a single bolus. Peak concentrations will greatly exceed those found on the standard regimen, but this does not seem to result in increased toxicity. What is important is that the pre-dose


concentration should be <1 mg/l, indicating complete elimination of the drug before another dose is given. Aminoglycoside assay should be done within 48 h of starting therapy and weekly thereafter, provided renal function is stable and levels satisfactory.

Assay of glycopeptides

Vancomycin and teicoplanin are increasingly used to treat infections with methi-cillin-resistant Staph. aureus (MRSA) and multi-resistant enterococci. These drugs have long plasma half-lives and may accumulate in patients with renal impairment. For this reason, and because of the reputation of vancomycin for toxicity, assay is often done. However, commercial preparations of vancomycin are now of a much higher purity and are less toxic than earlier ones (which earned the nickname ‘Mississippi mud’), and the necessity of monitoring plasma levels has been questioned.

There is, however, evidence that high peak concentrations of glycopeptides are associated with improved outcome and that prolonged trough concentrations may relate to nephrotoxicity. The target therapeutic range should be 5–15 mg/l (pre-dose) and 20–40 mg/l (1 h post-dose) for vancomycin; 10–15 mg (pre-dose) and >40 mg/l (post-dose) for teicoplanin. Assays should be done for all patients receiving more than 4 days’ treatment, and performed weekly if satisfactory levels are achieved.

Assay of other antibiotics

There are few reasons in routine clinical practice to monitor the concentrations of antimicrobial drugs other than aminoglycosides and glycopeptides. Although chloramphenicol is no longer widely used to treat meningitis in neonates, monitoring of plasma and CSF levels is necessary if this is contemplated, since these infants are at risk of developing the ‘grey baby syndrome’ because of their inability to metabolize the drug.

Antibiotic assay methods

Numerous methods have been described to assay antimicrobial substances. Some compounds, such as β-lactam antibiotics, lend themselves to chemical assay methods, but these are often insufficiently sensitive to measure the small amounts that may be present in body fluids and in some cases do not discriminate between active compound and inactivate metabolites. Commercially available immunoas-says have now virtually replaced other methods for the assay of aminoglycosides and other antibiotics in laboratories where the cost of the machine and reagents is not prohibitive.



Microbiological assay

Bacteriological methods directly quantify antibacterial activity. The main problems with these assays are the slow turn-round time (often requiring overnight incubation) and their low precision.

The simplest method is to titrate the antibiotic-containing fluid against a known sensitive organism and to compare the result with a parallel titration of a standard solution of the drug in question. The standard solution should be initially prepared in a similar fluid to that of the test (serum, urine, etc.). Although the method has the virtue of simplicity it is not very accurate because test and standard are being compared in a discontinuous series of concentrations and because antibiotic titrations are inherently irreproducible within twofold limits.

More satisfactory is the agar diffusion method of assay. In this test a series of wells is cut in a prearranged pattern in agar contained in a flat-bottomed plate. The agar may be seeded with the test organism before distribution in dishes, so that the bacteria are in the agar, or the surface may be flooded with an appropriately diluted suspension of bacteria and allowed to dry before the wells are cut. The wells are filled with the test fluid (which may be diluted if high levels are anticipated) and a series of standard antibiotic concentrations prepared in the same body fluid as the test.

After overnight incubation at 37 °C the diameter of each inhibition zone is carefully measured (Fig. 8.5). A graph of the square of the zone diameter is plotted against the logarithm of the antibiotic concentration for the antibiotic standards. A linear relationship should be obtained if the test is working satisfactorily. The levels of antibiotic in the test fluid can then be determined by reference to the graph (Fig. 8.6).


Fig. 8.5 Microbiological assay of antibiotic levels in patient's serum. Wells cut in agar (previously seeded with an indicator organism) are filled with patient's serum, or standard dilutions of antibiotic of known potency. After overnight incubation the zones of inhibition are measured, a graph is drawn from results obtained with the standards (see Fig. 8.6) and the concentration in the patient's serum is read off from the graph. In the example shown, the figures represent the concentrations of the antibiotic standards used; patient's serum was obtained 1 h after receiving a dose of antibiotic (peak level, P) and immediately before administration of the next dose (trough level, T). (Photograph courtesy of George Sharp and Richard Edwards, Nottingham Public Health Laboratory.)


Fig. 8.6 Calibration graph used for estimating concentrations of an antibiotic in patient's serum. Values obtained with standard concentrations of the antibiotic in a test such as that depicted in Fig. 8.5 are plotted as shown. The concentration of antibiotic in the appropriate test sample can then be read off by reference to the graph. In the example shown, the patient's serum produced an inhibition zone measuring 22.5 mm in diameter (zone squared = 506 mm2) indicating an antibiotic concentration of 5.1 mg/l. (Graph, based on an assay of gentamicin in serum, kindly provided by Anthony Cowlishaw, Nottingham Public Health Laboratory.)

The organism used in the assay should not be so sensitive to the antibiotic in question that huge inhibition zones are produced by therapeutic concentrations. Bacillus subtilisSarcina lutea, or the Oxford strain of Staph. aureus are widely used. For gentamicin assay done by the microbiological method it is commonplace to use a fast-growing organism such as Klebsiella aerogenes so that results can be obtained within 4–5 h if required. If a strain of Klebsiella is chosen which has a restricted susceptibility pattern, interference with the assay by other antibacterial agents can be minimized. Since the method merely detects antimicrobial activity, antibiotics other than the one being assayed may interfere with the result unless steps are taken to inactivate the other compound, or to use an organism that is resistant to it. It is essential that assay requests made to the laboratory should reveal all antimicrobial therapy that the patient is currently receiving.


Various kinds of immunoassay have been described. They are fast and accurate, but expensive on reagents and equipment. Antibody against the antibiotic, prepared in rabbits, is allowed to react with the patient's serum where it combines with any antibiotic that is present. A known amount of fluorescein-labelled or


enzyme-labelled antibiotic is then added to the mixture and reacts with any remaining antibody. The fluorescence or enzyme activity is measured and the loss, compared to a control value, is proportional to the amount of antibiotic in the serum. The reagents for this test must be highly standardized and a fluo-rimeter or spectrophotometer is required to read the result. The methods in most common use are the fluorescence polarization method (TDX) and the enzyme-multiplied immunoassay technique (EMIT).

High-pressure liquid chromatography (HPLC)

This is a development of traditional chromatographic procedures. Compounds are separated according to their differential retention times as they are passed under


pressure through a special column containing particles which delay the test substance to a greater or lesser degree depending on small variations in surface charge. By manipulating the conditions of the fluid (mobile phase) in which the test material is dissolved, the system can be made highly selective. Substances which have a characteristic ultraviolet (UV) absorption spectrum can be detected rapidly and quantitatively by continuously monitoring the outflow of the chromatographic column by UV spectrophotometry. Compounds which do not absorb UV can usually be detected in other ways, often by chemically treating the sample to form a fluorescent derivative. This must be done for aminoglycosides and this, together with the extraction procedure needed to remove proteins and other interfering substances from serum samples, makes the method too cumbersome for routine use.

However, HPLC is a useful and versatile method for the assay of many antimicrobial agents in pharmacokinetic studies. It has the great advantage that it


can discriminate between very closely related compounds and can therefore be used to detect not only native antibiotic, but also any derivatives that may be produced in the body, which may display modified pharmacological, toxicological, or antibacterial properties.