The aim of the present study was to propose specific critical inhibition diameters for the Nocardia genus that could accurately discriminate between susceptible and non-susceptible strains to the main antibiotics used in the treatment of nocardiosis.

The CLSI warns of the risk of false resistance to ceftriaxone for N. cyriacigeorgica and N. brasiliensis species, as well as for the risk of false resistance to imipenem for N. farcinica and N. cyriacigeorgica species, notably due to an extension of the incubation period for DST to 72 h [10]. Regarding the incubation time for broth microdilution DST, the extension to 72 h had a strong impact on MIC values and consecutively on interpretative categories for imipenem for N. farcinica and N. cyriacigeorgica, as well as for ceftriaxone for N. cyriacigeorgica; with the risk of producing false R results. We therefore strongly advise laboratories performing broth microdilution DST for Nocardia to systematically read the plates for these species after a 48 h incubation. On the other hand, for N. brasiliensis, and despite our limited number of strains (n = 8), we found that a 48 h incubation did not sufficiently detect the expected resistance to imipenem (I or R were detected only in 75% of cases after 48 h incubation), whereas an extension to 72 h incubation had no significant impact on ceftriaxone results. For this reason, we would recommend to read the broth microdilution DST of N. brasiliensis after a 72 h incubation.

For disk diffusion DST, despite a high discriminatory power after the analysis of the AUC of the ROC curve for several antibiotics (imipenem and amoxicillin-clavulanate after 48 h incubation, and amikacin, tobramycin, and fluoroquinolones after 48 h and 72 h incubation), the FDA criteria to assess the performance acceptability of disk diffusion method were reached only for tobramycin after a 72 h incubation (critical inhibition diameter of 19 mm). Moreover, for tobramycin at 72 h incubation, given that 1/2 strains categorized S using MIC determination in broth microdilution with inhibition diameters < 19 mm belong to the N. nova species, expected to be resistant [13], the determination of the interpretative category for this strain would be more reliable using the disk diffusion DST than the broth microdilution DST. Considering that the true interpretative category of these strains was amikacin R (according to their species identification), then we would obtain a CA of 99.3% 95%CI [95.9%; 99.9%], a vmj rate 0% 95%CI [0.0%; 5.4%] and a maj rate 1.5% [0.0%; 7.9%] (data not shown). Interestingly, the critical inhibition diameter of 19 mm is almost exactly the median of the range that was approximately fifty years ago proposed by Wallace et al. for a 10 µg tobramycin disk load: R ≤ 15 mm, S ≥ 25 S [9].

For the other conditions, it might be interesting to investigate in details the results for which the FDA criteria were close to be reached, e.g. for amikacin 72 h and for imipenem 72 h, excluding the data for N. farcinica and N. cyriacigeorgica. For amikacin, after 72 h of incubation, where the estimated vmj rate was zero and the maj rate was < 3%, a critical diameter of 25 mm appears to accurately discriminate between susceptible and non-susceptible (I or R) Nocardia strains. This is especially true considering that the two strains categorized as susceptible based on MIC determination by broth microdilution, with inhibition diameters < 25 mm, belong to the N. wallacei species, which is expected to be resistant [13]. In this case, interpretative category determination would be more reliable using the disk diffusion DST than the broth microdilution DST. Considering that the true interpretative category of these strains was amikacin R (according to their species identification), then we would obtain a vmj rate 0% 95%CI [0.0%; 45.9%] and a maj rate 0% [0.0%; 2.8%] (data not shown). The upper limit of 95%CI of vmj rate would remain > 7.5%. Nevertheless, in order to obtain an upper limit of the 95%CI < 7.5% for an estimated vmj rate of 0%, a minimum of 48 amikacin-R Nocardia strains would have to be included [15]. Amikacin resistance in Nocardia is unusual, apart from the N. transvalensis complex [16], and this species is not the most frequently found in human pathology [6, 16,17,18], thus, enriching the collection with N. transvalensis complex strains would appear to be a complicated task. Interestingly, this critical inhibition diameter of 25 mm is exactly the median of the range that was proposed approximately fifty years ago by Wallace et al. for a 30 µg amikacin disk load: R ≤ 20 mm, S ≥ 30 mm [9].

For imipenem at 72 h, excluding the data for N. farcinica and N. cyriacigeorgica, for which the estimated vmj rate was 0.0% and the maj rate < 3%, a critical diameter of 29 mm seems to accurately discriminate between susceptible and non-susceptible (I or R) Nocardia strains. The upper limit of 95%CI of vmj rate was > 7.5% due to insufficient included number of strains categorized imipenem R (n = 26). In order to obtain an upper limit of the 95%CI < 7.5% for an estimated vmj rate of 0%, a minimum of 48 imipenem-R Nocardia strains would have to be included. As an imipenem resistance is more common in Nocardia strains, it would be possible to complete the available data with additional imipenem-R strains in order to confirm this result. Since the vast majority (61/62) of the N. farcinica and N. cyriacigeorgica strains retained a diameter > 29 mm around the imipenem disk even after 72 h incubation, this confirms that the low discriminating power observed at 72 h for imipenem, when considering all Nocardia strains, was due to these strains, categorized as falsely R to imipenem with higher MIC due to the extended 72 h delay. For these species, extending the incubation period from 48 h to 72 h seems to have less impact on inhibition diameters around the imipenem disk than on imipenem microdilution MIC determination. For N. farcinica and N. cyriacigeorgica, admitting the clinical categorization S, I, and R obtained through the MIC for imipenem at 48 h incubation (available for 55 strains) and considering a critical inhibition diameter of 29 mm for imipenem after 72 h incubation, the CA would be 90.9% 95%CI [80.0; 96.9] but two N. farcinica strains would remain with vmj discrepancies (strains categorized R with imipenem MIC at 48 h incubation and S with imipenem disk diffusion DST at 72 h incubation), one N. cyriacigeorgica strain with maj discrepancies (strain categorized S with imipenem MIC at 48 h incubation and R with imipenem disk diffusion DST at 72 h incubation), and two N. farcinica strains with a minor discrepancy (strains categorized I with imipenem MIC at 48 h incubation and categorized S with imipenem disk diffusion DST at 72 h incubation; data not shown). Hence, even if the critical inhibition diameter for imipenem at 72 h incubation of 29 mm was confirmed, it would only be valid for Nocardia species other than N. cyriacigeorgica and N. farcinica. Yet these two species are the most frequently found in human pathology [6, 16,17,18] and imipenem is one of the drugs of choice in the probabilistic treatment of Nocardiosis [2]. Consequently, although limited, the interest in having disk diffusion DST validated exclusively for Nocardia species other than N. farcinica and N. cyriacigeorgica, especially for such an important drug, must be considered.

For amoxicillin-clavulanate after a 48 h incubation, the determination of the interpretative category could be more reliable by using a critical inhibition diameter around the disk of 20 mm than the broth microdilution DST. Considering that the true interpretative category was amoxicillin-clavulanate-R for the 9 strains having MIC ≤ 8/4 mg/L and an amoxicillin-clavulanate inhibition diameters ≤ 20, and according to species identification (N. nova complex and N. cyriacigeorgica,, which were expected to be R to amoxicillin-clavulanate according to CLSI [13]), it would allow to improve the maj rate (0.0% 95%CI [0.0%; 7.4]; data not shown) and the CA (97.3% 95%CI [92.4; 99.4], data not shown) without allowing the vmj rate (2.6% 95%CI [0.0; 13.5], data not shown) to reach the FDA criteria. Herein, the determination of a zone I for amoxicillin-clavulanate diffusion DST would not enable to improve the criteria since the strain with vmj discrepancies and a MIC > 64/32 mg/L had an inhibition diameter around the amoxicillin-clavulanate disk (30 mm) too distant from the critical inhibition diameter (20 mm).

For other antibiotics with an AUC of the ROC curves > 0.9 (imipenem and fluoroquinolones), determining a zone I for disk diffusion DST did not meet the FDA criteria. This was partly due to the strains with vmj discrepancies for which the inhibition diameters were widely distributed toward higher values.

Although the composition of MH-F medium is strictly controlled, and that it is the most appropriate medium for cotrimoxazole susceptibility testing for fastidious germs such as Haemophilus influenzae [19], it seems unsuitable to test this antibiotic on Nocardia; in the present study, 25% of the strains showed a contact inhibition diameter to the cotrimoxazole disk (6 mm), while broth microdilution DST categorized all Nocardia strains as susceptible. In the study by Wallace et al. in which a critical diameter of 20 mm was proposed for cotrimoxazole 1.25–23.75 µg, they found 94% of susceptible strains [8]. This implies that they did not observe strains with contact bacterial diameters, which may indicate that MH-F medium is not suitable for cotrimoxazole disk diffusion DST on Nocardia. This finding may also be due to the difficult reading of the diameter around the cotrimoxazole disk, which often includes a zone of partial inhibition, making the 80% inhibition reading subjective. Nevertheless, the choice of MH-F medium for the study is questionable. While it provides certainty regarding the growth of all Nocardia species, even the most fastidious (e.g. N. nova complex, N. abscessus complex), it should be noted that some species are not particularly fastidious (N. farcinica and N. cyriacigeorgica) and can easily grow on conventional MH medium. It would then be interesting to see whether a change of medium could improve the performance of the disk diffusion DST and to determine specific critical diameters for the MH medium.

Regarding the impact of Nocardia species on the DST technique, the present study did not reveal any particular species with high rates of vmj and maj discrepancies except for imipenem, for which all strains with vmj discrepancies were either N. farcinica or N. cyriacigeorgica after 48 h and 72 h of incubation. Therefore there is no particular species that could not be tested for DST using disk diffusion. However, to precisely assess this point, experiments would have to be carried out on large collections containing a single species of Nocardia, which may be difficult to obtain for particularly rare species.