With regard to antibiotic susceptibility profiles, the present study shows that the number of antimicrobials that can be used in probabilistic antibiotic therapy with guaranteed effectiveness in cases of nocardiosis is low (LZD, AMI, SXT), that no beta-lactams have a high susceptibility rate (> 90%) against Nocardia, and that species identification is essential for selecting the best antimicrobial among beta-lactams. It is of note that FOX and FEP were clearly unsuitable for the treatment of nocardiosis (0.8% and 5.3% of susceptible isolates, and Figure S3), and therefore these 2 antimicrobials can be omitted from the plate for Nocardia AST. For other cephalosporins, MIC50 for AXO and FOT were close, as were the MIC90, for all categories of Nocardia tested (N. cyriacigeorgica, N. farcinica, N. nova complex and all Nocardia). However, the frequency of FOT-S isolates was drastically lower than that of AXO-S (Figure S4). This was due to the use of different reference standards for interpreting clinical categories (CLSI critical concentrations for Nocardia spp and other aerobic actinomycetes [23], 2018, for AXO and PK/PD breakpoint of CA-SFM 2025 for FOT [24]). This highlights the major impact of the thresholds defined by different learned societies on the interpretation of MIC and the determination of interpretative clinical categories, highlighting the need to evolve these thresholds as knowledge progresses.

Concerning carbapenems, the MIC data found herein were also consistent with that reported for N. farcinica, showing that IMI was better than MERO: 77.4% of S/I isolates versus 22.6%, in the present study; 63% of S/I isolates versus 33% previously published by Brown-Elliott et al. [17]; and 92.3% of S/I isolates versus 84.7% previously published by Yang et al. [18]. Moreover, MERO MIC may be better than those of IMI for certain Nocardia species such as N. wallacei [17]. However, IMI was also shown slightly better than MERO for N. cyriacigeorgica in the present study (93.9% of S/I isolates versus 81.8%), contrary to that previously published by Brown-Elliott et al. [17] and Yang et al. [18] (60% of S/I isolates versus 68%; and 94.45% of S/I isolates versus 97.2%, respectively). In the present study, other species such as N. gamkensis and N. gipuzkoensis displayed a MIC for MERO 2 dilutions lower than that for IMI. Thus, on a case-by-case basis, MERO could be an alternative for IMI-R Nocardia strains, and determining its MIC is necessary in order to prescribe this antibiotic to patients. Although no MERO-specific breakpoint for the genus Nocardia is available for interpreting clinical categories, PK/PD breakpoints recommended by the CA-SFM 2025 (S ≤ 2 mg/L, I = 4 and 8 mg/L, R ≥ 16 mg/L [24]), used herein, or bacterial interpretative thresholds of the CLSI (S ≤ 4 mg/L, I = 8 mg/L, R ≥ 16 mg/L [17, 18]), used by Brown-Elliott et al. and Yang et al., could be used to guide clinicians regarding its potential use. Although these breakpoints differ slightly, the CA-SFM definition of I = S, increased exposure, would allow the use of MERO against Nocardia isolates with a MIC up to 8 mg/L.

Concerning TZD, the MIC was lower than that found for LZD, which is consistent with the literature [8]. There are currently no thresholds defined by learned societies (CLSI, EUCAST or CA-SFM) for determining clinical categories for TZD and Nocardia. Recently, a study evaluated a PK/PD threshold for TZD and LZD and established it at 0.5 and 1 µg/mL, respectively [30]. Applying these breakpoints in the present study, the frequency of Nocardia strains susceptible to TZD appears have a much higher than that susceptible to LZD (93.2% versus 29.3%, respectively, data not shown). Furthermore, the PK/PD breakpoint of 1 µg/mL for LZD, which has been adopted by CA-SFM 2025 [24], raises questions about the CLSI’s current breakpoint of 8 µg/mL [23] and suggests that it should be updated. In the present study, one Nocardia strain belonging to N. brasiliensis exhibited higher MIC to oxazolidinones than the others (TZD 2 µg/mL and LZD 8 µg/mL). It may be worthwhile to sequence this strain to determine whether genetic determinants could explain this reduced susceptibility.

Regarding the comparison of the two AST plates used herein, all FDA criteria (EA, CA, ME, and the upper/lower limit of the 95% CI of the VME rate) were respected for 6 antimicrobials (AUG2, AXO, CIP, CLA, MXF, and TOB) on the 13 antimicrobials contained in both plates. For LZD, the VME rate could not be calculated due to the absence of LZD-resistant Nocardia strains, as expected by the recommendations of the CLSI [23]; otherwise, the CA, EA and ME rate criteria were met. The enrichment of the collection with LZD-resistant strains appears to be very difficult, especially when considering the CLSI’s breakpoint.

For 5 antimicrobials (AMI, DOX, IMI, MIN and SXT), the only unmet FDA criterion was the upper limit of 95% CI of VME rate, which exceeded > 7.5%, despite a point estimate and a lower limit of the 95% CI of 0.0%. Achieving an upper limit of the 95% CI < 7.5% for an estimated VME rate of 0.0% would require at least 48 resistant Nocardia strains per antimicrobial [25]. For IMI and DOX, this target was nearly reached, and an enrichment with 10 IMI-R strains and 4 DOX-R strains could be considered. However, this appears to be a complicated task for the remaining antimicrobials. For AMI, resistance in Nocardia is uncommon, except in the N. transvalensis complex [26], which is not commonly involved in human disease [26,27,28,29]. In our collection, only 6 N. wallacei (belonging to the N. transvalensis complex) were identified, including 3 with AMI MIC > 8 mg/L and 2 categorized as S to AMI with an MIC of 8 mg/L. Although this value remains within the S range [23], it is markedly higher than the MIC90 (< 1 mg/L) observed for most other Nocardia species. This suggests that AMI resistance in the N. transvalensis complex may be difficult be detected by microdilution AST even after 72-hour of incubation. Similarly, SXT resistance was uncommon (4.5%) and consistent with previous studies using microdilution AST technique, reporting resistance rates between 0.9% in a Chinese collection [26] and 6.5% in a Israeli collection [31]. The resistant isolates involved the same species previously reported, namely N. farcinica, N. wallacei and N. otitidiscavarium [26, 31, 32]. For MIN, the Chinese and Israeli studies mentioned above report, as herein, a high proportion of MIN-I strains with MIC of 2–4 µg/mL but very few MIN-R strains (0.9% and 0.0% respectively [26, 31]).

Another point of note is that, for TGC, only the EA was met. The reason for this was that the breakpoint used to define the clinical categories (S ≤ 0.5 µg/mL, R > 0.5 µg/mL) also corresponds to the modal MIC of all the strains tested, which divided the overall Nocardia population into two almost equal parts between TGC-S (46.6%) and TGC-R (53.4%). As no zone “I” was determined, it is very easy to move from S to R in one MIC dilution, resulting in poor CA, as well as ME and VME rates.

With regard to reproducibility testing, the tests carried out herein do not meet FDA requirements as they were conducted in a single center on only two strains of different species (S. aureus and N. asteroides). However, for internal use at the LBMR des nocardioses, these tests appear sufficient to ensure reproducible use of the device within the laboratory.

The main limitations of the present study, other than the lack of resistant strains for certain antimicrobials as mentioned above, were, first, that the FDA criteria could not assessed according to Nocardia species stratification because of the limited number of isolates for all species. Second, the reference method used to evaluate the FRNOCAR1 plate was a commercially available ready-to-use plate, RAPMYCOI. Nevertheless, the RAPMYCOI plate is IVD-labeled and has been evaluated and validated by the manufacturer for Nocardia spp., its use ensures a certain degree of accuracy of the results, although systematic biases may still be present. Third, certain antimicrobials, such as TZD, MERO, and FOT, were available only on the FRNOCAR1 plate, preventing comparison of the results with a reference method.

Regarding the latter point, the TZD MICs obtained with the FRNOCAR1 for the reference strain S. aureus ATCC 29,213 were within the expected target ranges 100% of the time. This result provides at least partial validation of the accuracy and of the reproducibility of the TZD MIC generated by the FRNOCAR1. Although no specific clinical breakpoint has been established for Nocardia spp. with TZD, preliminary PK/PD data [30] and reported clinical cases [9,10,11,12,13] suggest that TZD may represent an alternative treatment option for nocardiosis in selected situations such as resistance, adverse effects, or drug-drug interactions. Consequently, MIC values for this antimicrobial could be reported without clinical category interpretation, together with a comment recommending expert consultation prior to its use for the curative treatment of nocardiosis.

For MERO and FOT, the MIC obtained for the reference strain S. aureus ATCC 29,213, although reproducible, did not allow conclusions to be drawn regarding the accuracy of the MICs generated by the FRNOCAR1, in the absence of expected MIC ranges established by learned societies (CLSI or CA-SFM). The use of another quality control strain with well-established expected MIC ranges could be considered, such as Escherichia coli ATCC 25,922. However, because the expected MIC ranges for this strain are very low—0.03 to 0.125 mg/L for FOT and 0.008 to 0.06 mg/L for MERO—the concentration range available on the FRNOCAR1 would not allow for sufficiently accurate evaluation of these MIC. MERO already appears to be commonly used in certain countries, particularly China and India, especially for the treatment of cerebral nocardiosis [19,20,21]. In this regard, a recent meta-analysis on cerebral nocardiosis reported that MERO was combined with SXT for the antimicrobial treatment of these infections [33]. In this context, and although additional analyses remain necessary to ensure the accuracy of MERO MIC obtained using the FRNOCAR1, it appears desirable to be able to provide, under specific conditions, a MERO MIC determined by microdilution when available. This could help generate clinical efficacy data when MERO is used for treatment. Consequently, MERO MIC obtained with the FRNOCAR1 could be reported without clinical category interpretation, only upon justified request from an expert clinician, together with a comment indicating that the result is pending further validation for accuracy, interpretative criteria, and clinical correlation.

For FOT, the MICs were very similar to those of AXO, not offering any clear advantage over AXO for the treatment of nocardiosis. Consequently, the FOT MICs obtained with the FRNOCAR1 plate may remain unreported and be used for research purposes only.

In conclusion, for AMI, AUG2, AXO, CLA, CIP, DOX, IMI, LZD, MIN, MXF, SXT, and TOB, the FRNOCAR1 demonstrated good agreement compared with RAPMYCOI for Nocardia strains. Furthermore, the FRNOCAR1 plate enables the determination of TZD MIC, a parameter not available with existing ready-to-use commercial plates such as the RAPMYCOI. The LBMR des nocardioses therefore routinely uses the FRNOCAR1 for microdilution AST of Nocardia for the following agents: AMI, AUG2, AXO, CLA, CIP, DOX, IMI, LZD, MIN, MXF, SXT, TOB, and TZD. The MERO MIC data generated concurrently could be reported to clinicians upon justified request, accompanied by a comment stating that the results remain provisional pending full validation. The generated FOT MIC data are intended exclusively for research use. This approach allows results to be obtained within a timeframe compatible with clinical management and for antibiotics of therapeutic interest.