The SLOMYCO panel showed excellent performance for clarithromycin. However, a key limitation of our study is the absence of clarithromycin-resistant MAC isolates. Therefore, although our data clearly demonstrate the accuracy of the SLOMYCO panel for susceptible strains, additional studies including resistant isolates are needed for a more comprehensive evaluation. Nevertheless, the high CA and reproducibility observed here strongly support the use of the SLOMYCO system for routine clarithromycin susceptibility testing of MAC, in alignment with current clinical guidelines recommending macrolide susceptibility testing for all MAC cases. Clarithromycin susceptibility testing is clinically the most important assay for MAC, as macrolides remain the cornerstone of MAC therapy [4,12]. Therefore, this high level of agreement is a reassuring result. Our findings are consistent with those of Babady et al. [7], who reported 93% agreement between the Sensititre panel and the reference broth microdilution method for clarithromycin MICs with no categorical interpretation errors. In that study, the SLOMYCO panel was described as a simple and reliable testing system, and our results confirmed that even more than a decade later, its accuracy for clarithromycin testing remains robust. 

Notably, the distribution of the clarithromycin MICs differed between the two methods. In the reference method (KIT), 21 of 86 isolates (24.4%) showed MIC values ≤ 0.5 µg/mL, which is higher than typically reported in the literature. For comparison, a previous study reported that the most frequent clarithromycin MIC values for MAC were 1 and 2 µg/mL [13]. In our SLOMYCO results, the MIC distribution was shifted: only 2 isolates (2.3%) had MIC ≤ 0.5 µg/mL, while 45 isolates (52.3%) had MIC = 2 µg/mL, which is more consistent with previous reports. It is difficult to determine which method reflects the true MIC more accurately, as only a limited number of laboratories worldwide continue to perform conventional broth microdilution, whereas commercial systems such as SLOMYCO have become widely adopted. Given that the manufacturing process of commercial panels is standardized and undergoes quality control, inter-laboratory variation may be lower than that of reagents prepared in-house. This discrepancy in MIC distribution represents a limitation of our study and highlights the need for further investigation of the factors influencing MIC determination in different testing systems.

Our evaluation suggested that the SLOMYCO panel is reasonably reliable for amikacin testing, albeit with minor caveats. We observed 90% EA between amikacin MICs and 77.9% categorical agreement. However, a key limitation of this study was the absence of amikacin-resistant MAC isolates. Thus, while our data demonstrated good reliability for susceptible or intermediate strains, further studies including resistant isolates are needed for a more complete evaluation. Despite this limitation, the fact that only one major error (false R) occurred is encouraging.

Most observed discrepancies were minor errors, mainly involving intermediate-range MICs (32 mg/L). For example, one isolate showed an MIC of 32 mg/L by the reference method (technically “intermediate”) but 16 mg/L by SLOMYCO (susceptible), and another showed the opposite pattern. These borderline results are well recognized in all systems. Indeed, Jaffré et al. [14] reported that using Middlebrook 7H9 broth instead of Mueller-Hinton broth for MAC DST reduced the frequency of intermediate results for amikacin. They noted that the MIC distribution of amikacin in MAC tended to cluster around the breakpoint, making the interpretation highly sensitive to small methodological variations. They also emphasized that the amikacin breakpoint (16 mg/L) is one of the few breakpoints that can be used “rationally” for MAC, but differences in media or methodology can shift an isolate from susceptible to intermediate.

Our findings reflect this issue; slight differences in growth or reading led to approximately 20% of isolates being categorized differently (S vs. I) between the methods. Clinically, whether the MIC is 16 or 32 mg/L may not be decisive; the drug can often still be used, although the latter may require higher dosing or yield only a partial response. Nevertheless, the relatively high minor error rate suggests that laboratories should exercise caution when reporting “intermediate” amikacin results. It would be prudent to retest the borderline isolates or at least mention the uncertainty associated with intermediate MICs.

Overall, our results support the conclusion that the SLOMYCO panel is a practical and reliable tool for testing amikacin susceptibility in MAC. Given that recent clinical guidelines recommend baseline amikacin susceptibility testing when a drug is included in a regimen, our findings provide further evidence that the panel method can effectively fulfill this role [12,15].

Fluoroquinolones are not first-line drugs for MAC, as studies have not shown clear efficacy; however, moxifloxacin is sometimes used in salvage regimens, and ciprofloxacin has historically been used in MACdisseminated disease in AIDS patients [4,12]. We found a modest CA of 69.8% for moxifloxacin and a very high minor error rate (30.2%). Essentially all discrepancies were S↔I or R↔I differences. Notably, our data did not show any instances where one method called an isolate susceptible and the other resistant (which would be a major discrepancy); instead, they disagreed on whether an isolate with an MIC of 2–4 mg/L should be deemed intermediate or resistant. This reflects the fact that the current tentative breakpoints for moxifloxacin (S ≤ 1, I = 2, R ≥ 4) may be problematic. Indeed, Zimenkov et al. [16] reviewed multiple studies and noted that only a few had MIC distributions that cleanly supported these breakpoints, and many laboratories found that most MAC isolates had MICs of 1–2 mg/L for moxifloxacin. In this scenario, any small MIC variation toggles the interpretation of S and I. Our observation that SLOMYCO often yielded one dilution higher MIC than the reference (e.g., turning an I into an R) may be due to slight differences in medium composition or the fact that the panel’s moxifloxacin range starts at 0.12 mg/L (with doubling dilutions) whereas some laboratories might use different increments. However, given the lack of a proven clinical utility of moxifloxacin in MAC, these minor discrepancies may not be clinically significant. However, in the laboratory, susceptibility to moxifloxacin can be inconsistent. One approach is to report moxifloxacin not as susceptible/resistant but simply provide the MIC with the comment that its clinical role is unclear.

We also observed a poor EA (67%) for ciprofloxacin MICs between the methods, although we did not have breakpoints to compare the categories. This parallels the findings from a study on M. marinum, where ciprofloxacin had a high EA (~98%), but doxycycline and ethionamide did not [17]. In our MAC isolates, ciprofloxacin MICs were generally higher than those of moxifloxacin (most often 2–8 mg/L), and because there are no MAC breakpoints for ciprofloxacin, these differences are mostly of academic interest. Given that neither ciprofloxacin nor moxifloxacin is routinely recommended for MAC disease, the discrepancies here, although interesting, may not affect patient management in most cases.

The most striking result of our study was the very low CA for linezolid (47.7%). This is not entirely surprising; the efficacy of linezolid against MAC is uncertain, and breakpoints have only recently been suggested, largely based on wild-type MIC distributions and limited clinical data. Jaffré et al. [14] noted that the current linezolid breakpoints for MAC (S ≤ 8, R ≥ 32) cut through the wild-type distribution, resulting in a “moderate” kappa agreement at best between test methods. In their media comparison, they found that the MIC modal values were the same in 7H9 and MH broth for linezolid; however, when converted to susceptible/resistant categories, the agreement was only moderate, implying that many isolates fell near the breakpoints. Our study confirms that even within one medium (presumably, both methods use a similar broth base), linezolid MICs for MAC often hover around 16–32 mg/L. Thus, whether an isolate is labeled as intermediate or resistant depends on a single dilution difference, which is common between the SLOMYCO and manual methods. The high minor error rate (51.2%) indicated that approximately half of the isolates were interpreted differently (I vs. R) using the two methods. What do these discrepancies imply? From a clinical perspective, because linezolid is not a standard first-line MAC drug, clinicians may only use it in salvage therapy when options are limited. In such cases, an intermediate MIC (16 mg/L) is difficult to interpret, and some experts may still attempt the drug (at a higher dose or in combination) if there are few alternatives. The key is that the laboratory must communicate with uncertainty. Our findings suggest that if one were to use the SLOMYCO panel for linezolid testing, it would reliably identify clearly susceptible strains (MIC ≤ 8) versus clearly resistant (≥ 32), but the intermediate range will yield inconsistent results. Laboratories could consider reporting linezolid MIC values rather than categorical results, or if reporting categories, note that reproducibility around the breakpoint is poor. Further research should refine these breakpoints or identify more reproducible testing conditions. It is worth noting that a Brazilian study by Garcia Carvalho et al. [8] found higher CA for linezolid when using a colorimetric read (RAPMYCO panel for rapid growers had linezolid 100% agreement, and the SLOMYCO linezolid agreement was not separately reported, but the overall SLOMYCO agreement for the included drugs was 60%–80%). Although not directly comparable, this suggests that methodological tweaks (such as adding resazurin dye for a clearer endpoint) could improve the consistency for drugs such as linezolid.

In our study, ethambutol demonstrated the highest EA among all tested drugs, reaching 98% between the SLOMYCO panel and the reference CLSI broth microdilution method. This indicated a strong level of reproducibility and consistency between the two testing systems. The MIC values for ethambutol clustered tightly and only a few isolates showed one dilution differences between the methods. However, the interpretation of ethambutol MICs for MAC remains complex, as no CLSI categorical breakpoints are currently established for MAC. This limitation prevents the direct comparison of CA and the determination of error rates. Although ethambutol is routinely used as a standard MAC treatment regimen (typically in combination with macrolides and rifamycins), its true clinical efficacy against MAC remains uncertain, and the correlation between MIC values and treatment outcomes has not been clearly established [4,12].

We observed a low EA between doxycycline and TMP-SMX. These agents are generally not part of MAC therapy (although minocycline or doxycycline might have some activity against certain NTM, and TMPSMX is used for M. fortuitum and others, but not MAC). Discrepancies in MICs likely stem from difficulties in in vitro testing. Sulphonamides often produce trailing endpoints, and tetracyclines may have partial inhibitory effects in vitro. Indeed, in the M. marinum SLOMYCO evaluation, doxycycline had the lowest EAs (72%), and trimethoprim (tested separately from sulfamethoxazole) had an EA of 74%, which is very close to what we found in MAC. This pattern of high agreement for some drugs and much lower agreement for others seems intrinsic to the behavior of antibiotics in the broth. Thus, the results for doxycycline and TMP-SMX against MAC should be interpreted with caution. As there were no breakpoints, one could argue that there was no clinical need to test them in MAC. We included them mainly because they were on the commercial panel; their presence did not hinder the test, but laboratories should be careful not to overinterpret these MIC values. Some laboratories may choose not to report certain drug results from the panel, if they are not clinically relevant.