Origin of the test organism

Although two cultures may bear the same generic and specific name, i.e. they may both be called Escherichia coli, this does not mean that they are identical. Certainly, they would normally be similar in many respects, e.g. morphology (appearance), cultural requirements and biochemical characteristics, but they may exhibit slight variations in some of these properties; such variants are described as strains of E. coli. A variety of strains of a single species may normally be obtained from a culture collection, e.g. the National Collection of Industrial and Marine Bacteria or the National Collection of Type Cultures. Different strains may also occur in hospital pathology laboratories by isolation from swabs taken from infected patients or by isolation from contaminated food, cosmetic or pharmaceutical products, and from many other sources. Strains obtained in these ways are likely to exhibit variations in resistance to antimicrobial chemicals. Strains from human or animal infections are frequently more resistant to antimicrobial chemicals, particularly antibiotics, than those from other sources. Similarly, strains derived from contaminated medicines may be more resistant to preservative chemicals than those obtained from culture collections. Therefore, in order to achieve results that are reproducible by a variety of laboratories, it is necessary to specify the strain of the organism used for the determination.

It is becoming increasingly common, too, for official testing methods to limit the number of times the culture collection specimen may be re-grown in fresh medium (called the number of subcultures or passages) before it must be replaced. This is because the characteristics of the organism (including its resistance to antimicrobial chemicals) may progressively change as a result of mutation and natural selection through the many generations that might arise during months or years of laboratory cultivation.

Composition and pH of the culture medium

There are several methods of assessing antimicrobial activity which all have in common the measurement of inhibition of growth of a test organism when the antimicrobial chemical is added to the culture medium. In such cases the composition and pH of the medium may influence the result. The medium may contain substances that antagonize the action of the test compound, e.g. high concentrations of thymidine or para-aminobenzoic acid will interfere with sulfonamide activity.

The antimicrobial activities of several groups of chemical are influenced by the ease with which they cross the cell membrane and interfere with the metabolism of the cell. This, in turn, is influenced by the lipid solubility of the substance, because the membrane contains a high proportion of lipid and tends to permit the passage of lipid-soluble substances. Many antimicrobial chemicals are weak acids or weak bases, which are more lipid soluble in the unionized form. The pH of the environment therefore affects their degree of ionization, hence their lipid solubility and so, ultimately, their antimicrobial effect. Benzoic acid, for example, is a preservative used in several oral mixtures which has a much greater activity in liquids buffered to an acid pH value than those which are neutral or alkaline. Conversely, the aminoglycoside antibiotics, e.g. streptomycin, neomycin and gentamicin, which are weak bases, are more active at slightly alkaline pH values, although this is more a consequence of the transport systems by which the molecules enter the bacterial cell working better at alkaline pH than of enhanced lipid solubility. The presence of organic matter, e.g. blood, pus or serum, is likely to have a marked protective effect on the test organism and so antimicrobial chemicals may appear less active in the presence of such material. The activity of several antibiotics, notably tetracyclines and aminoglycosides, is reduced by the presence of high concentrations of di- or trivalent cations in the medium.

Exposure and incubation conditions

The temperature, duration and redox conditions of exposure to the antimicrobial chemical (or incubation of survivors after exposure) may all have a significant effect on its measured activity. Increasing the temperature of exposure of the test organism to the chemical increases the antimicrobial activity by a factor which is quantified by the temperature coefficient (Q₁₀ value: the number of times increase in activity for a 10 °C rise in temperature). Phenols and alcohols, for example, may respectively exhibit Q₁₀ values of 3–5 and > 10, and so a variation of 5 °C in the temperature of exposure (which is permitted by pharmacopoeial preservative efficacy tests) may lead to a markedly different rate of kill of the organism in question.

The period of time for which the test organism is exposed to the antimicrobial chemical may influence the recorded result because it is possible for the organism to adapt and become resistant to the presence of the chemical. In preservative efficacy tests, the exposure period is normally 28 days, which is sufficient time for any cells that are not killed during the first 24–48 hours to recover and start to reproduce, so that the final bacterial concentration may be much higher than that at the start. This is illustrated in Figure 14.1, which shows the effect of the quaternary ammonium preservative benzethonium chloride on Pseudomonas aeruginosa. The concentration of bacteria was reduced to approximately 0.01% of the initial value during the first 6 hours, but the bacteria that survived this early period recovered to the original level within 2 days. There is the potential for a similar phenomenon to arise in other situations, e.g. in minimum inhibitory concentration (MIC) determinations of bacteriostatic agents (those that do not kill but merely inhibit the growth of the test organism), although it is not common in MICs because the exposure (incubation) time is much shorter than that in preservative testing.

The effect of some antibiotics may be influenced by the redox conditions during their period of contact with the test organism. Aminoglycosides, for example, are far less active, and metronidazole is far more active, under conditions of low oxygen availability. Such effects may even be seen during agar diffusion antibiotic assays, in which the antibiotic diffuses from a well into an agar gel inoculated with the test organism; the diameter of the zone of growth inhibition that surrounds a well filled with neomycin solution, for example, may be significantly greater at the surface of the agar (where there is abundant oxygen) than at its base, where the oxygen concentration is limited by its poor diffusion through the gel.

Inoculum concentration and physiological state

It is perhaps not surprising that the concentration of the inoculum can markedly affect antimicrobial action, with high inoculum levels tending to result in reduced activity. There are two main reasons for this. First, there is the phenomenon of drug adsorption on to the cell surface or absorption into the interior of the cell. If the number of drug molecules in the test tube is fixed yet the number of cells present is increased, this obviously results in fewer molecules available per cell and consequently the possibility of a diminished effect. In addition to this there is the second, more specialized case, again concerning antibiotics, where it is frequently observed that certain species of bacteria can synthesize antibiotic-inactivating enzymes, the most common of which are the various types of β-lactamases (those destroying penicillin, cephalosporin and related antibiotics). Thus a high inoculum means a high carryover of enzyme with the inoculum cells, or at least a greater potential synthetic capacity.

Perhaps less predictable than the inoculum concentration effect is the possibility of the inoculum history influencing the result. There is a substantial amount of evidence to show that the manner in which the inoculum of the test organism has been grown and prepared can significantly influence its susceptibility to toxic chemicals. Features such as the nature of the culture medium, e.g. nutrient broth or a defined glucose-salts medium, the metal ion composition of the medium and hence of the cells themselves, and the physiological state of the cells, i.e. ‘young’ actively growing cells from the logarithmic growth phase or ‘old’ non-dividing cells from the stationary phase, all have the potential to influence the observed experimental values. Generally, antimicrobial chemicals are more effective against actively growing cells than slowly growing or dormant ones, e.g. bacterial spores.

• conventional chemical assays, e.g. titrations, spectrophotometry and high-performance liquid chromatography (HPLC)

• enzyme-based and immunoassays, where the antibiotic is, respectively, the substrate for a specific enzyme or the antigen with which a specific antibody combines

• biological assays in which biological activity, in this case bacterial growth inhibition, of the ‘test’ solution is compared with that of a reference standard.

• when the antibiotic is present in a solution containing a wide variety of complex substances that would interfere with a chemical assay, e.g. fermentation broth, serum, or urine

• when the antibiotic is present together with significant concentrations of its breakdown products, e.g. during stability studies as part of product development

• when it has been extracted from a formulated medicine, for example a cream or linctus, when excipients might cause interference

• where the commercially available product is a mixture of isomers that have inherently different antimicrobial activities, which cannot easily be distinguished chemically and which may differ in proportion from batch to batch (e.g. neomycin and gentamicin).

Agar diffusion assays

In this technique, the agar medium in a Petri dish or a larger assay plate is inoculated with the test organism, wells are created by removing circular plugs of agar, and these wells are filled with a solution of the chemical under test (Fig. 14.2).

The chemical diffuses through the gel from A towards B and the concentration falls steadily in that direction. The concentration in the region A to X is sufficiently high to prevent growth, i.e. it is an inhibitory concentration. Between X and B the concentration is sub-inhibitory and growth occurs. The concentration at X at the time the zone edge is formed is known as the critical inhibitory concentration (CIC). After incubation, the gel between A and X is clear and that between X and B is opaque as a result of microbial growth which, with the common test organisms, is usually profuse. A zone of inhibition is therefore created, the diameter of which will increase as the concentration of chemical in the well increases.

A graph may be constructed which relates zone diameter to the logarithm of the concentration of the solution in the well (Fig. 14.3). It is normally found to be linear over a small concentration range but the square of the diameter must be plotted to achieve linearity over a wide range. A plot such as that in Figure 14.3 may, quite correctly, be used to calculate the concentration of a test solution of antibiotic. In practice, however, it is found to be more convenient to obtain reliable mean zone diameters for the standard at just two or three concentrations, rather than somewhat less reliable values for six or seven concentrations. There is no reason why an assay should not be based upon a two- or three-point line, provided that those points are reliable and that preliminary experiments have shown that the plotted relationship over the concentration range in question is linear.

It is not common to conduct antibiotic assays in Petri dishes because too few zones may be accommodated on a standard-sized dish to permit the replication necessary to obtain the required accuracy and precision. Antibiotic assays, when performed on a large scale, are more often conducted using large assay plates 300 mm or more square (Fig. 14.4). The wells are created in a square design and the number that may be accommodated will depend upon the anticipated zone diameters: 36 or 64 wells are common (6 × 6 or 8 × 8, respectively). The antibiotic standard material may be used in solution at three known concentrations (frequently referred to as ‘doses’) and the antibiotic solution of unknown concentration treated likewise; alternatively, each may be employed at two concentrations. A randomization pattern known as a Latin square is used to ensure that there is a suitable distribution of the solutions over the plate, thereby minimizing any errors due to uneven agar thickness.

In the case of an assay based upon standard solutions used at two concentrations, the potency ratio may be calculated directly from the graph (as shown in Fig. 14.5) or by using the formula below:

(14.1)

where X is the potency ratio, LDR is the logarithm of the dose ratio (i.e. ratio of concentrations of standard solutions), UH, UL, SH and SL are the mean zone diameters for the unknown and standard high and low doses. The derivation of this is described in detail by Wardlaw (1999), who deals extensively with the subject of antibiotic assays. The tests for acceptable limits of parallelism between the line joining the standards and that joining the test points, together with confidence limits applicable to the calculated potency ratios, are described in the current PhEur.

In calculating the potency ratio directly from Figure 14.5, the zone diameters for the standard and unknown high concentrations are plotted at the same abscissa values, and those for the low concentrations similarly. Two zone diameters are considered which are as widely separated on the ordinate as possible while still being covered by the standard and the test lines. The ratio of the concentrations required to achieve the selected diameter is thus an estimate of the potency ratio. The mean of the two estimates taken at the extremes of the range of common zone diameters should be identical to the value by calculation from the formula. Thus, in Figure 14.5, at a zone diameter of 23.75 mm the first estimate of potency ratio is 0.557 (antilog of 0.445 divided by antilog of 0.699); the second is 0.507 (antilog of zero divided by the antilog of 0.295). The mean value of 0.53 indicates the unknown solution to have approximately half the activity of the standard.

Turbidimetric assays

In this case, antibiotic standards at several concentrations are incorporated into liquid media and the extent of growth inhibition of the test organism is measured turbidimetrically using a nephelometer or spectrophotometer. The unknown or test antibiotic preparation is run simultaneously, again at several concentrations, and the degree of growth inhibition compared. Such assays are less commonly used than agar diffusion methods because their precision is rather inferior but they do have the advantage of speed: the result may be available after an incubation period as short as 3–4 hours. They are also more sensitive than diffusion assays and consequently may be applied to low-activity preparations.

The shape and slope of the dose–response plot for a turbidimetric assay may be more variable than that for agar diffusion, and non-linear plots are common. Typical dose–response plots are shown in Hewitt & Vincent (1989). The plotted points are usually the mean turbidity values obtained from replicate tubes and the assay may be conducted using a Latin square arrangement of tubes incubated in a shaker, which is necessary to ensure adequate aeration and uniform growth throughout the tube.

MIC test methods

The most common way to conduct MIC determinations is to incorporate the antimicrobial chemical at a range of concentrations into a liquid medium, the containers of which are then inoculated, incubated and examined for growth.

Test tubes may be used, but microtitre plates (small rectangular plastic trays with, usually, 96 wells each holding approximately 0.1 mL liquid) and other miniaturized systems are common. It is also possible to incorporate the chemical into molten agar, which is then poured into Petri dishes and allowed to set. An advantage of using a microtitre plate or series of Petri dishes is that several organisms can be tested at the same time using a multipoint inoculator; there is also a greater chance of detecting contaminating organisms (as uncharacteristic colonies) on the agar surface than in liquid media. Usually the presence or absence of growth is easier to distinguish on the surface of agar than in liquid media. In tubes showing only faint turbidity, it is often difficult to decide whether growth has occurred or not. Regardless of the method used, the principle is the same and the MIC is the lowest concentration at which growth is inhibited.

In addition to the other experimental details that should be described in order to make the measured result meaningful, it is necessary to specify the increment by which the concentration of test chemical changes from one container to the next. The operator could, for example, change the concentration 10-fold from one tube to the next in the rare circumstance where even the likely order of magnitude of the MIC is not known. Far more commonly, however, the concentration changes by a factor of 2, and this is almost invariably the case when antibiotic MIC values are determined; thus, reference is made to ‘doubling dilutions’ of the antibiotic. If, for example, an MIC were to be measured using test tubes, an aqueous solution of the chemical would normally be mixed with an equal volume of double-strength growth medium in the first tube in the series, then half the contents of the first tube added to an equal volume of single-strength medium in the second, and so on. In this case half the contents of the last tube in the series would have to be discarded prior to inoculation in order to maintain the same volume in each tube. Control tubes may be included to demonstrate (a) that the inoculum culture was viable and that the medium was suitable for its growth (a tube containing medium and inoculum but no test chemical) and (b) that the operator was not contaminating the tubes with other organisms during preparation (a tube with no test chemical or added inoculum). It is possible to use an arithmetic series of concentrations of test chemical, e.g. 0.1, 0.2, 0.3, 0.4, … rather than 0.1, 0.2, 0.4, 0.8, … µg mL⁻¹. The potential problem with this approach is that there may be merely a gradation in growth inhibition, rather than a sharp point of demarcation with obvious growth in one tube in the series and no growth in the next.

All the solutions used must be sterilized; it must not be assumed that the test chemical is self-sterilizing. Most disinfectants, antiseptic and preservative chemicals are bactericidal but they are unlikely to kill bacterial spores. Also, several antibiotics act by inhibiting growth and so would not necessarily kill vegetative cells with which they might be contaminated. If the experiment is conducted in tubes, all the tube contents must be mixed before inoculation as well as after, otherwise there is the possibility of the inoculum cells being killed by an artificially high concentration of the test chemical towards the top of the tube. If there is any risk of precipitation of the test chemical or the medium components during incubation, a turbidity comparison must be available for each concentration (same tube contents without inoculum); alternatively, in the case of bactericidal chemicals, the liquid in each tube may be subcultured into pure medium to see whether the inoculum has survived. Each of the tubes in the series may be prepared in duplicate or triplicate if it is considered desirable. This is the case where the incremental change in concentration is small.

Distinction between MICs conducted in agar and the assessment of sensitivity using agar diffusion methods

It is important to understand that when MICs are determined in Petri dishes the antimicrobial chemical is dissolved in the agar and is uniformly distributed through the gel when the test organism is inoculated into the surface. This is a fundamental difference from the test procedure used for antibiotic bioassays where the antibiotic diffuses through the agar to create a growth inhibition zone. When MICs are conducted in agar there is no diffusion and no zones of growth inhibition; the result merely depends on the presence or absence of growth of the test organism.

If the agar diffusion method were used to measure the size of the inhibition zones from a series of solutions of progressively decreasing concentration, it would obviously be possible to identify the concentration that just fails to produce an inhibition zone. This is sometimes incorrectly described as the MIC value for the chemical in question; such a procedure, however, gives the critical inhibitory concentration (CIC), not the MIC. CIC values usually exceed MIC values by a factor of 2–4. Not only is this misconception about agar diffusion methods giving MIC values commonly found in the pharmaceutical and chemical literature but misinterpretations of agar diffusion data are, unfortunately, also common. The diameter of a growth inhibition zone depends upon several factors. Whilst the sensitivity of the test organism, its concentration and that of the chemical are paramount, the incubation conditions, the physicochemical composition of the gelled culture medium and the properties of the diffusing molecule are also important. It is tempting to take the simplistic view that if two chemicals are used at the same concentration and one produces a larger zone of growth inhibition than the other, that is a direct reflection of their intrinsic antimicrobial activities. Unfortunately, that is often not the case because it fails to take into account both the diffusion coefficients of the different molecules and their concentration exponents (Chapter 15). To diffuse well in agar, a molecule should be small, water soluble and of a charge that does not interact with the components of the gel. There are several very effective antimicrobial chemicals that either do not diffuse well in agar or possess a high concentration exponent, both of which properties would predispose to small zones. If these agents were to be assessed purely on the basis of inhibition zone diameter they would be incorrectly dismissed as virtually inactive. Parabens and phenols are prime examples. Even saturated solutions of parabens in water can fail to give inhibition zones by agar diffusion (Fig. 14.6) but they are, nevertheless, amongst the most effective and widely used antimicrobial preservatives. Likewise phenols, with their high concentration exponents, only give small inhibition zones and this has led to misleading comparisons; manuka honey, for example, has been claimed to possess antibacterial activity equivalent to 10% phenol on the basis that the inhibition zone diameters are similar. This fundamental limitation of agar diffusion as a method of assessing antimicrobial potency is all too frequently overlooked.

Choice of test organisms and inoculum concentration

The test organisms used are the bacteria Staphylococcus aureus, Pseudomonas aeruginosa and E. coli (which is used for testing all product types in the USP test but for oral products only in the PhEur test), together with the yeasts/moulds Candida albicans and Aspergillus brasiliensis (plus the osmophilic Zygosaccharomyces rouxii in the PhEur test for oral syrups). The current PhEur recommends that the designated organisms be supplemented, where appropriate, by other strains or species that may represent likely contaminants to the preparation. A similar recommendation was contained in earlier versions of the USP preservative test but this has been deleted from the current test (2010).

One problem with adding other organisms (such as those isolated from the manufacturing environment) is that they are not universally available and so a particular product could be tested at different manufacturing sites of the same company and pass in one location yet fail in another simply because the organisms used locally were not the same. The possibility of using resistant strains isolated from previous batches of spoilt product has been advocated, but this too may pose problems because organisms may rapidly lose their preservative resistance unless routinely grown on media supplemented with the preservative in question.

The inoculum concentration of 10⁵–10⁶ microorganisms mL⁻¹ or g⁻¹ of the preparation under test has been criticized as being unrealistic because it is much higher than that which would be acceptable in a freshly manufactured product. It is adopted, however, in order for the 1000-fold fall in microbial concentration that would be required from an effective parenteral or ophthalmic preservative to be easily measured. The test organisms are added separately to different containers rather than as a mixed inoculum.

Inactivation of preservative

It is quite possible for sufficient of the preservative to be contained in, and carried over with, the sample removed from the container to prevent or retard growth of colonies on the Petri dishes. If the inoculum level of the test organism initially is about 10⁶ cells mL⁻¹ or g⁻¹ of product, the problem of carryover may not arise because a dilution factor of 10³ or 10⁴ would be required to achieve a countable number of colonies on a plate; at this dilution most preservatives would no longer be active. When a high proportion of the cells in the product has died, however, little or no such dilution is required, so preservative carryover is a real problem which may artificially depress the count even more. To avoid this, preservative inhibitors or antagonists may be used. There are several of these; common examples being glycine for aldehydes, thioglycollate or cysteine for heavy metals, and mixtures of lecithin and polysorbate-80 with or without Lubrol W for quaternary ammonium compounds, chlorhexidine and parabens. The use of these and other inactivators has been tabulated by Gilmore et al (2011).

An alternative method of removing residual preservative is to pass the sample of inoculated product through a bacteria-proof membrane so that surviving organisms are retained and washed on the surface of the membrane and the preservative is thus physically separated from them. After washing, the membrane is transferred to the surface of a suitable agar medium and colonies of microorganisms develop on it in the normal way. It is necessary to incorporate controls (validate the method) to demonstrate both that the inactivator really works and that it is not, itself, toxic. The former usually involves mixing the inactivator with the concentrations of preservative likely to be carried over, then inoculating and demonstrating no viability loss. Details of these validation procedures are described more fully in chapter <1227> of the USP (2010).

One further control is a viable count of the inoculum performed by dilution in peptone water to check the actual number of cells introduced into the product. This is necessary because even a ‘zero time sample’ of the product will contain cells that have been exposed to the preservative for a short period as it usually takes 15–45 seconds or more to mix the inoculum with the product and then remove the sample. Some of the cells may be killed even in such a short time and so a viable count of the inoculum culture will reflect this.

Interpretation of results

The extent of microbial killing required at the various sampling times for a preservative to be considered acceptable for use in parenteral or ophthalmic products is greater than that required for a preservative to be used in topical products, which in turn exceeds that for an oral product preservative (Table 14.2).

In the case of the first two product categories, the PhEur specifies two alternative performance criteria, designated A and B. The A criteria express the recommended efficacy to be achieved, whereas the B criteria must be satisfied in justified cases where the A criteria cannot be attained, for example because of an increased risk of adverse reactions. The baseline used as the reference point to assess the extent of killing is the concentration of microorganisms expected to arise in the product after addition and mixing of the inoculum, as calculated from a viable count performed on the concentrated inoculum suspension prior to its addition to the product. The viable count of the time-zero samples removed from the inoculated product is not the baseline.

• suspension tests

• capacity tests which measure the extent to which the disinfectant can withstand repeated additions of test organisms

• carrier tests, where the organism is loaded or dried on to a carrier

• in-use tests, which are intended to simulate actual conditions of use as closely as possible.

1. Because disinfectants are normally used in circumstances where there is a significant amount of organic ‘dirt’ present, modern testing procedures invariably attempt to take this into consideration. Thus, yeast, albumin or other material is added in known concentration to the disinfectant/microorganism mixture.

2. Regardless of the method by which the antimicrobial activity is assessed (see below), it is a fundamental principle of disinfectant testing, just as it is with preservative efficacy tests, that the antimicrobial activity of the disinfectant must be halted (also referred to as neutralized, inactivated or quenched) in the sample when it is removed from the disinfectant/organism mixture. Clearly, meaningful results cannot be obtained if it is impossible to distinguish what fraction of the microbial killing occurred during the timed period of exposure to the disinfectant from that arising due to carryover of disinfectant into the incubation step that follows exposure. Verification that the disinfectant inactivation method is effective and that any chemical neutralizers used are, themselves, non-toxic to the test organisms is an integral part of the test.

3. It is in viability assessment that there is a fundamental difference of approach between recently developed tests (exemplified by BS EN 1276) and many of the tests that originated before the 1980s. The simplest method of viability assessment, which was employed in the Rideal–Walker and Kelsey–Sykes tests, for example, is to transfer the sample from the disinfectant/microorganism mixture to a known volume of neutralizing broth, incubate and examine for growth (manifest as turbidity). This procedure contains the inherent defect that any growth in the tubes of broth may result from the transfer of very few surviving cells, or from many. Thus, it is possible for the disinfectant to kill a high proportion of the inoculum within a short period yet fail to kill a small fraction of the cells, possibly mutants, which have atypically high resistance. In this case, there is the risk that the disinfectant may be dismissed as insufficiently active despite the fact that it achieved a rapid and extensive initial kill. For this reason it has become common for disinfectant and preservative efficacy tests to be very similar in design, in that both employ viable counting methods to assess microorganism survival but the former utilize a sampling period of minutes or hours, whereas the latter use a 28-day period.

4. When viable counting is used to assess the survival of test organisms, the adoption of disinfection performance criteria based upon a required reduction in the number of surviving organisms is a logical strategy, just as it is in preservative testing. Thus, the so-called 5–5–5 testing principle has found much favour. Here, five test organisms are (separately) exposed for 5 minutes to the disinfectant, which is considered satisfactory if a 5-log reduction in viable numbers (a 10⁵ fall in viable cells mL⁻¹) is recorded in each case. This principle is adopted in the BS EN 1276, although only four bacterial strains are recommended for routine use; there is, however, the option to supplement the standard organisms with others more relevant to the intended use of the disinfectant in question.

Environmental monitoring

Environmental monitoring is normally taken to mean regular monitoring of the levels of microbial contamination of the atmosphere, of solid surfaces and, less frequently, of the personnel in the production areas. Water used to clean floors, benches and equipment (as distinct from water incorporated in the product) may be considered as part of environmental monitoring but will not be considered here as the procedures for counting microorganisms in water are described below.

Atmospheric monitoring is most commonly undertaken by means of settle plates, which are simply Petri dishes containing media suitable for the growth of bacteria and/or yeasts and moulds, e.g. tryptone soya agar, which are exposed to the atmosphere for periods of, typically, 1–4 hours. Microorganisms in the air may exist as single cells, e.g. mould spores, but more commonly they are attached to dust particles, so that any organisms in the latter category (for which the culture medium is suitable) will grow into visible colonies during incubation after dust particles have settled on the agar surface. The colony counts recorded on the plates are obviously influenced by:

The disadvantage of settle plates is that it is not possible to relate colony counts directly to air volume. This limitation is overcome in active sampling methods, whereby a known volume of air is drawn over, or caused to impact upon, the agar surface. These methods and the equipment available for active sampling have been reviewed by Johnson (2003).

Surface and equipment sampling is most frequently undertaken by swabbing or the use of contact plates (also known as RODAC – replicate organism detection and counting – plates). Swabbing a known area of bench, floor or equipment with a culture medium-soaked swab is convenient for irregular surfaces. The organisms on the swab may be counted after they have been dispersed by agitation into a fixed volume of suspending medium but it is not easy to quantify either the proportion of total organisms removed from the swabbed surface or the proportion dispersed in the diluent. This second limitation is overcome using contact plates, which are simply specially designed Petri dishes slightly overfilled with molten agar which, on setting, present a convex surface that projects above the rim of the plate. When the plate is inverted on to the surface to be sampled, microorganisms are transferred directly on to the agar (Fig. 14.7).

Sampling of manufacturing personnel usually consists of sampling clothing, face masks or, more commonly, gloves. ‘Finger dabs’ is the phrase used to describe the process whereby an operator rolls the gloved surface of each finger over a suitable solid medium in a manner similar to that in which fingerprints are taken. Operator sampling by any means other than finger dabs is rare, particularly outside aseptic manufacturing areas.

Counting of microorganisms in pharmaceutical products

Most pharmaceutical raw materials are contaminated with microorganisms. The levels of contamination are often a reflection of the source of the raw material in question, with ‘natural’ products derived from vegetable or animal sources, or mined minerals such as kaolin and talc, being more heavily contaminated than synthetic materials whose microbial burden has been reduced by heat, extremes of pH or organic solvents during the course of manufacture. Determining the bioburden in these materials is often straightforward, utilizing without modification the viable counting procedures described in Chapter 13. Occasionally the physical nature of the raw material makes this difficult or impossible, and this is often found to be the case with the finished manufactured medicine, where problems of dispersibility, sedimentation or viscosity cause complications. As a consequence, modifications to the standard viable counting procedures are necessary to reduce errors. Some of modifications and the circumstances that necessitate them are considered below.

Detection of specific hazardous organisms

In addition to placing limits on the maximum concentration of microorganisms that is acceptable in different materials, pharmacopoeias usually specify certain organisms that must not be present at all. In practice, this means that detection methods which are described in the pharmacopoeia must be applied to a known weight of material (typically 1–10 g), and the sample passes the test if, on the culture plates, no organisms arise that conform to the standard textbook descriptions of those to be excluded. Typically, the pharmacopoeial methods involve preliminary stages using selective liquid culture media; these are designed to increase the concentration of the organism that is the subject of the test (‘target’ organism) and so render it more readily detectable. Commercially available identification kits or specific supplementary biochemical tests may also be used to confirm the identity of any isolates having the typical appearance of the target organisms. The PhEur used to recommend appropriate supplementary tests but these have been removed from the current edition, not because of a lack of reliability but because identification kits have become more common.

Both the PhEur (2011) and the USP (2010) describe detection tests for Staphylococcus aureus, Pseudomonas aeruginosa, Escherichia coli, salmonellae and Candida albicans. In addition, the PhEur describes a test for clostridia, but this is unlikely to be applied to any material other than mined minerals, e.g. talc and bentonite. The five organisms common to both pharmacopoeias are the subject of these tests primarily because of their potential to cause infections. However, they may also represent common contaminants of the products to which the tests are applied, or their presence may be indicative of the quality of the raw material or finished manufactured product. E. coli, for example, is a natural inhabitant of mammalian intestines and so its presence in a material such as gelatin (which originates in the slaughterhouse) would indicate unacceptable quality. The most likely source of Staphylococcus aureus in a manufactured medicine is the production personnel, so that if this origin were confirmed it would indicate the need for higher manufacturing standards. In general, the tests are applied to pharmaceutical raw materials of ‘natural’ origin, e.g. carbohydrates, cellulose derivatives, gums and vegetable drugs. In addition, there is a requirement that products for use in the mouth, nose, ears or on the skin should be free of both Pseudomonas aeruginosa and Staphylococcus aureus and vaginal products should also be free from Candida albicans. Table 14.4 summarizes the PhEur (2011) testing schemes for the five principal organisms of interest. These schemes are described in more detail elsewhere, together with photographs of the typical appearance of the organisms in question (Hodges 2000).

Microbiological assays of B-group vitamins

Just as HPLC has become the favoured method of antibiotic assay, so too has it become the method of choice for assaying B-group vitamins. Turbidimetric assays are still occasionally used, however, for example when insurmountable problems arise in resolving the many peaks that might arise on an HPLC chromatogram from a multivitamin product (which may contain 10 or more active ingredients plus excipients, all of which may cause assay interference).

Microbiological assays of B-group vitamins employ similar techniques to turbidimetric assays of antibiotics (see earlier in this chapter). A culture medium is used which is suitable for the assay organism, except for the omission of the vitamin in question. The extent of bacterial growth in the medium is thus directly proportional to the amount of reference standard or test vitamin added. It is important to select an assay organism that has an absolute requirement for the substance in question and is unable to obtain it by metabolism of other medium components; species of Lactobacillus are often used for this purpose. ‘Carryover’ of the vitamin with the inoculum culture must be avoided because this results in some growth even when none of the test material has been added. Growth may be determined turbidimetrically or by acid production from sugars.

• the adoption of the highest possible specifications for the microbiological quality of the raw materials. The rationale here is that sterilization processes are more likely to be effective when the levels of microorganisms to be killed or removed (bioburdens) are as low as possible to begin with. Procedures used to determine bioburdens are described in Chapter 13 and earlier in this chapter

• the rigorous application of environmental monitoring procedures (as described above) during the course of manufacture, with more stringent limits for acceptable levels of microorganisms than those applicable during the manufacture of non-sterile products

• comprehensive validation procedures when sterilization processes are designed, together with regular in-process monitoring when those processes are in operation for product manufacture. Initial validation seeks to demonstrate that adequate sterilizing conditions are achieved throughout the load, and entails extensive testing with thermocouples, radiation dosimeters and biological indicators (see below) as appropriate.

Sterilization monitoring

Sterilization processes may be monitored physically, chemically or biologically (Denyer et al 2011). Physical methods are exemplified by thermocouples, which are routinely incorporated at different locations within an autoclave load, whereas chemical indicators usually exhibit a colour change after exposure to a heat sterilization process. Biological indicators consist of preparations of spores of the Bacillus species that exhibits the greatest degree of resistance to the sterilizing agent in question. The principle of their use is simply that if such spores are exposed to the sterilization process and fail to survive, it can be assumed that all other common organisms will also have been killed and the process is safe. Spores of Bacillus stearothermophilus (strictly speaking now Geobacillus stearothermophilus, although this name is still not in common use in the pharmaceutical literature) are used to monitor autoclaves and gaseous hydrogen peroxide or peracetic acid sterilization processes, whereas Bacillus atrophaeus is the organism normally employed for dry heat, ethylene oxide and low-temperature steam-formaldehyde methods; Bacillus pumilus is used in radiation sterilization procedures.

Such biological indicators are regularly employed for validation of a sterilization process which is under development for a new product, or when a new autoclave is being commissioned; they are not normally used for routine monitoring during product manufacture. Spores possess the advantage that they are relatively easy to produce, purify and dry on to an inert carrier, which is frequently an absorbent paper strip or disc, or a plastic or metal support. Spore resistance to the sterilizing agent must be carefully controlled and so rigorous standardization of production processes followed by observance of correct storage conditions and expiry dates are essential.

Tests for sterility

It is sufficient here to repeat that the test is really one for demonstrating the absence of gross contamination with readily grown microorganisms, and is not capable of affording a guarantee of sterility in any sample that passes.

The experimental details of these procedures are described in the PhEur (2011). This section is therefore restricted to an account of the major features of the test and a more detailed consideration of those practical aspects that are important or problematical.

It is obviously important that materials to be tested for sterility are not subject to contamination from the operator or the environment during the course of the test. For this reason it is essential that sterility tests are conducted in adequate laboratory facilities by competent and experienced personnel. Clearly, the consequences of recording an incorrect sterility result may be very severe. If a material which was really sterile were to fail the test it would need to be resterilized or, more probably, discarded. This would have significant cost implications. If, on the other hand, a contaminated batch were to pass a test for sterility and be released for use this would obviously represent a significant health hazard. For these reasons sterility testing procedures have improved significantly in recent years and failures are now viewed very seriously by the regulatory authorities. If a product does fail, it means either that the item in question is really contaminated, in which case the manufacturing procedures are seriously inadequate, or that the item is in fact sterile but the testing procedure is at fault. Either way, it is not possible to dismiss a failure lightly.

Sterility tests may be conducted in clean rooms or laminar flow cabinets which provide a grade A atmosphere as defined by the Rules and Guidance for Pharmaceutical Manufacturers and Distributors. However, it is becoming increasingly common for testing to be undertaken in an isolator that physically separates the operator from the test materials and so reduces the incidence of false-positive test results due to extraneous contamination introduced during the test itself. Such isolators are similar in principle to a glove box, and typically consist of a cabinet (supported on legs or a frame) that is sufficiently large for the operator, who is covered by a transparent hood of moulded flexible plastic forming the cabinet base, to sit or stand within it.

A sterility test may be conducted in two ways. The direct inoculation method involves the removal of samples from the product under test and their transfer to a range of culture media that might be expected to support the growth of contaminating organisms. After incubation the media are examined for evidence of growth which, if present, is taken to indicate that the product may not be sterile. It is not certain that the product is contaminated because the organisms responsible for the growth may have arisen from the operator or have been already present in the media to which the samples were transferred, i.e. the media used for the test were not themselves sterile. Thus, in conducting a sterility test it is necessary to include controls that indicate the likelihood of the contaminants arising from these sources; these are discussed below. The size and number of the samples to be taken are described in the PhEur (2011).

Again it is necessary to inactivate any antimicrobial substances contained in the sample. These may be the active drug, e.g. an antibiotic, or a preservative in an eye drop or multidose injection. Suitable inactivators may be added to the liquid test media to neutralize any antimicrobial substances, but in the case of antibiotics particularly, no such specific inactivators are available (with the exception of β-lactamases which hydrolyze penicillins and cephalosporins). This problem may be overcome using a membrane filtration technique. This alternative method of conducting sterility tests is obviously only applicable to aqueous or oily solutions that will pass through a membrane having a pore size sufficiently small to retain bacteria. The membrane, and hence the bacteria retained on it, is washed with isotonic salts solution, which should remove any last traces of antimicrobial substances. It is then placed in a suitable liquid culture medium. This method is certainly to be preferred to direct inoculation because there is a greater chance of effective neutralization of antimicrobial substances.

Solids may be dissolved in an appropriate solvent. This is almost invariably water because most other common solvents have antimicrobial activity. If no suitable solvent can be found the broth dilution method is the only one available. If there is no specific inactivator available for antimicrobial substances that may be present in the solid then their dilution to an ineffective concentration by use of a large volume of medium is the only course remaining.

The controls associated with a sterility test are particularly important because incomplete control of the test may lead to erroneous results. Failure to neutralize a preservative completely may lead to contaminants in the batch going undetected and subsequently initiating an infection when the product is introduced into the body.

The PhEur (2011) recommends that four controls are incorporated. The so-called growth promotion test simply involves the addition of low inocula (not more than 100 cells or spores per container) of suitable test organisms into the media used in the test to show that they do support the growth of the common contaminants for which they are intended. Staphylococcus aureus, Bacillus subtilis and Pseudomonas aeruginosa are the three aerobic bacteria used, Clostridium sporogenes the anaerobic bacterium and Candida albicans and Aspergillus brasiliensis the fungi. Organisms having particular nutritional requirements, such as blood, milk or serum, are not included, so they, in addition to the more obvious omissions such as viruses, cannot be detected in a routine sterility test because suitable cultural conditions are not provided. On the other hand, it is impossible to design an all-purpose medium, and sterilization processes that kill the spore-forming bacteria and other common contaminants are likely also to eradicate the more fastidious pathogens such as streptococci and Haemophilus species, which would be more readily detected on blood-containing media. This argument does not, however, cover the possibility of such pathogens entering the product, perhaps via defective seals or packaging, after the sterilization process itself and then going undetected in the sterility test.

The second control, termed the method suitability test, is intended to demonstrate that any preservative or antimicrobial substance has been effectively neutralized. This requires the addition of test organisms to containers of the various media as before but, in addition, samples of the material under test must also be added to give the same concentrations as those arising in the test itself. For the sterility test as a whole to be valid, growth must occur in each of the containers in these controls.

It is necessary also to incubate several tubes of the various media just as they are received by the operator. If the tubes are not opened but show signs of growth after incubation this is a clear indication that the medium is itself contaminated. This should be an extremely rare occurrence but, in view of the small additional cost or effort, the inclusion of such a control is worthwhile.

A control to check the likelihood of contamination being introduced during the test should be included in the programme of regular monitoring of test facilities. The PhEur 2011 recommends the use of ‘negative controls’, which may be employed to check the adequacy of facilities and operator technique. These items, identical to the sample to be tested, are manipulated in exactly the same way as the test samples. If, after incubation, there are signs of microbial growth in the media containing these negative controls, the conclusion is drawn that the contamination arose during the testing process itself.

Some items present particular difficulties in sterility testing because of their shape or size, e.g. surgical dressings and medical devices. These problems are most conveniently overcome simply by testing the whole sample rather than attempting to withdraw a portion of it. So, for example, large clear plastic bags which have been radiation sterilized may be used to hold the entire medical device or complete roll or pack of dressings, which would then be totally immersed in culture medium. This method would only be valid if the culture medium gained access to the entire sample; otherwise the possibility exists, for example, of aerobic bacterial spores trapped within it failing to grow owing to insufficient diffusion of oxygen. This approach has the advantage of imposing a more rigorous test because a much larger sample is used. In the case of dressings, it may also reduce the risk of operator-induced contamination compared to the alternative approach, which would require the withdrawal of representative samples for testing from different areas of the roll or pack.

The final aspect of the test which is worthy of comment is the interpretation of results. If there is evidence that any of the test samples are contaminated, the batch fails the test. If, however, there is convincing evidence that the test was invalid because the testing facility, procedure or media were inadequate, a single retest is permitted; this contrasts with earlier pharmacopoeial protocols, which under certain circumstances permitted two retests.

Endotoxin and pyrogen testing

This is an aspect of microbial contamination of medicines which is not normally considered part of microbiology but is discussed here because pyrogens are normally the products of microbial growth. A pyrogen is a material which when injected into a patient will cause a rise in body temperature (pyrexia). The lipopolysaccharides that comprise a major part of the cell wall of Gram-negative bacteria are called endotoxins, and it is these that are the most commonly encountered pyrogens (although any other substance that causes a rise in body temperature may be classified under the same heading). Bacterial cells may be pyrogenic even when they are dead and when they are fragmented, and so a solution or material that passes a test for sterility will not necessarily pass a pyrogen test. It follows from this that the more heavily contaminated with bacteria an aqueous injection becomes during manufacture, the more pyrogenic it is likely to be at the end of the process.

Two main procedures are used for the detection of pyrogens. The traditional method requires the administration of the sample to laboratory rabbits, whose body temperature is monitored for a period of time thereafter. The alternative procedure, which is now by far the most common, is to use the Limulus Amoebocyte Lysate Test (LAL), in which the pyrogen-containing sample causes gel formation in the lysis product of amoebocyte cells of the giant horseshoe crab Limulus polyphemus. A detailed account of endotoxin testing is outside the scope of this chapter but the review by Baines (2000) provides a comprehensive account of the practicalities of the method.