1. Introduction
The first published description of the Calibrated Dichotomous Susceptibility (CDS) Test appeared in “Pathology” in 19751. Diagnostic laboratories in Australasia soon adopted the CDS and it became the most commonly used method of susceptibility testing in this country. Some years ago, the CDS Users Group was formed and the feedback from this group stimulated and assisted in further development of the test. Since the original description of the CDS Test there have been, in addition to the published updates of the method, forty-two CDS Newsletters that have been distributed to members of the CDS Users Group.
Over time, several refinements have been introduced into the method and the scope of the CDS Test has been broadened to enable the vast majority of organisms encountered in a diagnostic laboratory to be tested using all of the currently available antimicrobials. Despite these changes the principles and requirements underlying the test remain the same. These include: firstly, the requirement that before any antibiotic can be tested by the method, it must be calibrated, that is, the size of the zones of inhibition observed with each species must be correlated with quantitative values (MIC) and secondly, that in the performance of the test, the operator must adhere closely to the method, as described, thereby reproducing the conditions that pertained at the time of calibration.
Whilst a section on quality assurance is included in the method, the operator should remember that the most effective single quality assurance measure is to follow the prescribed technique assiduously. There is no doubt that somebody will find a simpler or more effective way of performing one or more of the steps in the CDS Test. However, before any improvement can be incorporated into the method it is necessary to confirm that it does not disturb the correlation between zone size and MIC.
1.1. Unique features and basis of the CDS Test
Readers are referred to the original monograph on the CDS1 (reproduced on the CDS website), for a description of the theoretical basis of antibiotic disc testing. Some of the unique features of the CDS Test are described here and in the course of this, it is hoped that the derivation of the CDS name will become evident.
1.1.1. Calibration (C) of the Test
As the “gold standard” of the antibiotic susceptibility of an organism is the minimum inhibitory concentration (MIC) of the antibiotic under test, all the methods of susceptibility testing must relate to this value. Moreover, the MIC must be determined by an internationally standardised technique. Before any antibiotic or bacterial species is included in the CDS Test, the test must be calibrated for that particular antibiotic and the targeted species. Calibration consists of plotting the zone sizes observed with a large number of strains of the species included in the CDS Test against the log MIC of each antibiotic.
The agreed gold standard test is the agar dilution technique originally proposed by Ericsson and Sherris in 19712. The agar dilution method is still accepted by the WHO as the gold standard and it has considerable advantages over broth dilution, a technique that is used by other methods to determine the MIC. Some other methods of susceptibility testing have been using broth dilution as the quantitative technique to which the disc test is calibrated. This is a convenient technique that lends itself to greater automation and is supplied by commercial resources that have obtained an ISO 20776-1 standard for the method. We continue to calibrate the CDS test using agar dilution because it can be applied to a much broader range of organisms and does not suffer from the drawbacks associated with broth dilution that Ericsson and Sherris clearly demonstrated2.
A description of the agar dilution technique used to calibrate the CDS test is included as an essential part of the regulatory requirements set out by TGA for Australian laboratories using in-house IVDs (see page 21). It may also be of use to overseas users who may need to calibrate antibiotics that for one reason or another have not been calibrated by the CDS Reference Laboratory. For those laboratories who do not have access to a Steers’ replicator to inoculate the agar plates a satisfactory technique is to use a wire loop calibrated to deliver 4 bµL in a spot inoculation. The agar technique is described in detail below:
Determination of MIC by Agar Dilution (WHO)
Preparation of Media
Sensitest Agar
Sensitest Agar is reconstituted with distilled water according to the manufacturers’ instructions and steamed at 100°C to dissolve the agar-agar component. Forty-nine mL aliquots of the medium are dispensed into 100 mL capacity glass bottles with their caps loose and are autoclaved at 121°C for 15 minutes and then placed in a water bath at 50°C.
Sensitest Blood Agar
To 46.5 mL Sensitest Agar prepared as described above and cooled to 50°C in a water bath, add 2.5mL of defibrinated horse blood (final concentration of 5%).
Columbia Blood Agar Base chocolate agar
To 45 mL Columbia Blood Agar Base, prepared as described above and cooled to 70°C in a water bath, add 4 mL of defibrinated horse blood to achieve a final concentration of 8%. The contents are mixed and kept at 70°C for 15 minutes to obtain ‘chocolate agar’. The medium is then cooled to 50°C prior to use.
Haemophilus Test Medium
To 47 mL of Haemophilus Test Medium base (Oxoid CM898) sterilised and cooled to 50°C, add 1 mL of fresh or deep frozen of each of haematin and nicotinamide adenine dinucleotide (NAD) solutions to reach a final concentration of 15 mg/L of each. This medium has a shortened storage life and must be used within 2 weeks of preparation.
Supplemented Brucella Medium Base
To 44.5 mL Brucella Medium Base (Oxoid CM0169) sterilised and cooled to 50°C, add 2.5 mL of defibrinated horse blood (final concentration of 5%) and 1mL of a fresh or deep-frozen solution of haemin to a final concentration of 5 mg/L, and 1 mL of vitamin K to a final concentration of 1 mg/L.
Preparation of antibiotic solutions
Depending upon the antibiotic and the concentrations required, the antibiotic is weighed and dissolved in the appropriate solvent. The antibiotic solutions are sterilised using 0.45 µm Millipore filters and two-fold dilutions of the antibiotic are prepared in McCartney bottles using the appropriate sterile diluent. The antibiotic dilutions should cover the range appropriate for each antibiotic and for the organisms tested.
Preparation of antibiotic agar plates
After the medium has cooled to 50°C, a 1 mL volume of the antibiotic solution is added to a 49 mL aliquot of molten medium. The bottle is gently inverted to thoroughly mix the contents before dispensing 25 mL into each of the two Petri dishes which have been labelled with the appropriate antibiotic concentration. Plates may be used on the same day or stored at 4°C for one day. Prior to use, the plates are surface dried for 30 minutes at 35°C in an incubator.
Preparation of the inoculum
Bacterial strains are grown overnight on horse blood agar (chocolate agar is used for H. influenzae, Brucella agar is used for fastidious anaerobes) incubated at 35°C. Cell suspensions are prepared in sterile 0.9% saline (saline containing 20% peptone water is used for fastidious bacteria) and the turbidity of the suspensions is adjusted to 0.8 using a spectrophotometer set at a wavelength of 640 nm. These suspensions thus contain 109cfu/mL. Each suspension is diluted 100 fold using a standard 40 dropper Pasteur pipette that delivers one drop (0.025 mL) of the initial suspension into 2.5 mL diluent.
Replication of the inoculum
The two suspensions prepared as described above are dispensed into wells of a Steer’s replicator. The probes deliver 0.004 mL of the suspensions to the surface of the agar plates. For each organism, the inocula used contain 106 and 104 cfu. The control plate containing no antibiotic is inoculated first, followed by the remaining agar plates beginning at the lowest and progressing to the highest concentration of antibiotic. The inoculated plates are incubated in appropriate conditions for eighteen hours or longer, depending on the growth requirement of tested organisms. Appropriate reference strains are included as controls.
Reading of the antibiotic minimum inhibitory concentration (MIC)
The MIC is the concentration of antibiotic at which there is complete inhibition of growth of 104 inoculum except when the antibiotic is known to select resistant variants at a high frequency. A fine haze is ignored in the interpretation of the result.
1.1.2. Dichotomous (D) separation
The CDS Test divides and reports antibiotic susceptibilities simply into two categories, “susceptible” and “resistant”. We do not recognise “intermediate” as a valid category in the CDS Test. The reasons that we advanced early in the development of the test and which are still valid, were that, when it varied, the susceptibility of the common pathogens to the then available antibiotics was distributed bi‑modally. In those rare cases where some strains were less resistant than others we were able to demonstrate that no method of disc testing had sufficient precision to reliably define an “intermediate” group. Also in the present era of Evidence Based Medicine (EBM) the strongest case against reporting “intermediate” susceptibility is the dearth of evidence relating to the response to antibiotic therapy of infection caused by these isolates. As far as the CDS Test is concerned, they are classed as resistant because we regard the role of susceptibility testing is to act as a guide to the clinician in the choice of the most appropriate antibiotic.
1.1.3. Susceptibility (S) and break points
Over time we have adopted the term “susceptibility and susceptible” in preference to “sensitivity and sensitive” when these relate to CDS testing. The reason for this was the introduction of statistical analyses into CDS testing along with most other tests we perform in the clinical laboratory (see below). So as to avoid confusion between “antibiotic sensitivity” and “statistical sensitivity” we changed the former to “antibiotic susceptibility” and the categories of susceptibility to “resistant” and “susceptible”.
With many bacterial species if susceptibility to a particular antibiotic varies it naturally divides into one or two groups. In these cases the MICs are bi‑modally distributed into widely separated values and this is no problem in defining susceptible and resistant categories. With other species and particularly with many newer antibiotics the distribution of MIC’s is continuous and separation into categories of susceptibility is made on the basis of an arbitrary break point, irrespective of the method used.
Although there may be some supporting evidence such as clinical response, accepted tissue levels and extrapolation from experience and studies with closely related antibiotics in the majority of cases break points still are arbitrary values. The break point MICs of the CDS Test generally are similar to those of other methods. Where we do differ is that we tend to have a more conservative approach and we will select the lower end of the range of break point MICs as the CDS break point.
Even so, argument about a twofold difference in break points in different methods can only be considered as pseudo‑exactitude when it is remembered that the values are determined by a gold standard method of MIC determination, which uses doubling dilutions. Doubling dilution irrespective of the technique will have an inbuilt error in excess of the difference of the values under discussion.
1.1.4. Interpretation of results
Where possible the CDS Test uses a uniform zone size to define susceptible strains. The susceptible zone size of 6 mm annular radius (18 mm diameter) was not chosen at random but was that point of the diffusion sigmoid curve that enabled the greatest discrimination between susceptible and resistant strains with the majority of antibiotics having a similar diffusion constant. It is worthwhile revisiting here the Humphrey and Lightbown’s formula3 describing diffusion in agar that is reproduced in the original CDS monograph.
r2 = 9.21 Dt (logM – log 4πhDtc)
Where, for our purposes, r is the radius of the inhibitory zone, t is time from start, c is the MIC, D is the diffusion constant, M is the disc potency and h is depth of agar.
The simplest interpretation of this is that the zone size is directly proportional to the diffusion constant and the log of the disc potency and inversely proportional to the log of the MIC.
It can be seen from this formula that with antibiotics of a similar diffusion constant an appropriate adjustment of the disc potency with each antibiotic will result in isolates with different susceptible MIC to each of the antibiotics yielding a uniform zone size for all susceptible strains. On the other hand, if the diffusion constant of the antibiotic is markedly reduced and it is not possible to increase the disc potency, e.g., vancomycin and polymyxin, then the zone cut‑off point will need to be reduced.
Where one species has a susceptible MIC different from that of the predominant species when tested against a particular antibiotic, the designer of the test has two choices. Either the susceptible zone size can be adjusted, e.g., gentamicin with Pseudomonas species and the Enterobacteriaceae (4 mm), or the potency of the disc can be changed for that species alone, e.g., ampicillin with Haemophilus species versus the Enterobacteriaceae (5 μg v/s 25 μg).
1.1.5. Performance characteristics of the CDS Test
In common with other laboratory tests an assessment can be made of the performance characteristics of the CDS Test. Statistics such as sensitivity, specificity and the predictive value of the CDS Test can be calculated by relating the test results to those obtained with a standard quantitative method. In susceptibility testing statistical sensitivity measures how well the test correctly identifies “true susceptible” strains whereas specificity refers to the ability of the test to correctly categorise “true resistant” strains. The CDS Test is designed to achieve maximum specificity, i.e., the conditions of the test are set to avoid reporting a resistant strain as susceptible.
Laboratory tests rarely can achieve 100 per cent sensitivity and specificity. Similarly with the CDS Test it may be necessary to sacrifice some statistical sensitivity to achieve maximum specificity. In practical terms this means that with some calibrations a few marginally susceptible strains may not be correctly identified as such by susceptibility testing. With each calibration we also calculate the positive predictive value (PPV) of the test that measures the percentage of “true susceptibles” versus all susceptibles (true plus false) reported by the test. An acceptable calibration is one where the positive predictive value is over 98 per cent.
References
1 Bell, S.M. 1975. The CDS disc method of antibiotic sensitivity testing (Calibrated Dichotomous Sensitivity Test). Pathology. 7: Suppl 4, 1‑48.
2 Ericsson, H.M. & Sherris, J.C. 1971. Antibiotic Sensitivity Testing. Report of an international collaborative study. Acta Path. Microbiol. Scand B Microbiol Immunol. Suppl 217, 1-90.
3 Humphrey, J.H., & Lightbown, J.W. 1952. A general theory for plate assay of antibiotics with some practical applications. J. Gen. Microbiol. 7,129‑43.