Acinetobacter baumannii infections are associated with significant mortality, particularly in vulnerable ICU patients. Discordant results between the APEKS-NP and CREDIBLE-CR [5] studies and other case series [4, 6] have led to a reluctance to use cefiderocol to treat Acinetobacter baumannii complex infections. In two correspondences, Choby et al., 2021 [7, 8] suggested that these discrepancies might be due to cefiderocol-heteroresistant subpopulations not detected by standard AST. This hypothesis seems to be invalidated by the latest results from the CREDIBLE-CR study presented by Longshaw et al. (Poster 0423, ECCMID 2023). However, the apparently high rate of clinical or microbiological failure in some case series cannot be ignored, although other factors, such as clinical severity (septic shock) or strong immunosuppression, may influence these poor patient outcomes [6, 9]. Recently, the rapid development of adaptive resistance to this molecule in vitro was well documented by Stracquadanio et al. [10]. As mentioned in two recent reviews [11, 12], the clinical relevance of heteroresistance in the context of treatment failure has not been demonstrated with a sufficient level of evidence (retrospective studies, case reports) for any species other than S. aureus (meta-analyses). We prefer the term "adaptive resistance" over heteroresistance, as it is deliberately less precise (heteroresistance being defined as a subpopulation with a frequency ≤10− 6 [12]). Indeed, we wished to highlight the intrinsic capacity of this bacterial genus to adapt rapidly to this new drug in analogy with the notion of a “genetic barrier” in virology (HIV). Moreover, we used a technique simpler and more accessible than the current reference methods (PAPs, [8]), which cannot quantify the fraction of the subpopulation displaying ARC. We focused on the presence/absence of ARC in the isolates tested, to simulate as closely as possible the situation encountered in vivo and to orient potential use by clinicians. We also propose an alternative technique more easily reproduced by non-referent laboratories using commercial reagents (UMIC, Bruker daltonics).
We observed a clear clinical improvement in the patient studied here following combination therapy with cefiderocol and colistin. It should be noted that, despite the susceptibility of the isolate to cefiderocol in the standard AST performed at the time of diagnosis (four days after diagnosis), the ICU clinician did not select this treatment option, even though the microbiologist advised the use of a ꞵ-lactam, because of the results for the Acinetobacter subgroup obtained in the CREDIBLE-CR study. The two key points to note in this case are that (i) the isolate responsible for the infection did not develop ARC in vitro and (ii) the patient was treated with combination therapy, in this case with colimycin, rather than monotherapy.
Our microbiological analyses with this new protocol (adapted from that described by Weiss et al.) yielded very similar results to recent studies based on the reference technique (PAPs): 84.2% of the OXA-23 carbapenemase-producing CRAB isolates initially categorised as susceptible, 80% if we include the OXA-23 + PER-7 isolate initially categorised as resistant, developed ARC (80% in the study by Stracqadanio et al., [10]). For OXA-24-producing CRAB, it is difficult to draw any firm conclusions due to the very small number of isolates (ARC was detected in two of the four isolates). ARC developed in 69.2% of the isolates initially classified as susceptible in standard AST, considered together. Isolates producing another class D carbapenemase, OXA-58, did not develop ARC. All the isolates producing the NDM metallo-β-lactamase or PER extended-spectrum β-lactamase were initially categorised as resistant to cefiderocol in standard AST, as previously described [13]. Thus, the type of carbapenemase type can potentially guide the choice of treatment, with a presumptive prediction of microbiological category (NDM: R, OXA-23/24: HR, OXA-58: S). These preliminary results, of course, require confirmation in studies on larger numbers of isolates, but they are nevertheless highly relevant in the context of the development of new rapid and accessible lateral flow assays (Coris bioConcept, NG biotech) allowing the rapid characterisation of the carbapenemase type in an Acinetobacter spp. isolate. With effective clinical-biological dialogue, these rapid tests can provide clinicians with an indication of the risk of treatment failure on cefiderocol or the need to use combination therapy within a few minutes.
Heteroresistance is not a new phenomenon in microbiology, but its clinical implications remain unclear (except for S. aureus). It has been described for many antibiotics and combination treatments are generally used to prevent it [14]. Moreover, the toned to simulate iron depletion in vitro, as such depletion is thought to occur in vivo, should lead to caution in the interpretation of these microbiological results and the in vitro/in vivo correlation. Our findings suggest that cefiderocol should be considered as an alternative treatment for CRAB infections if the isolate is classified as susceptible to this antibiotic in standard AST and no other ꞵ-lactam is available. However, it should be used in combinations (colymicin, tigecycline or avibactam/sulbactam) to prevent the development of adaptive resistance phenomena [9, 10]. Finally, it may be useful to advise microbiologists to repeat standard AST (MIC) on Acinetobacter spp. isolates after 48 or 72 h of exposure in vivo to cefiderocol in cases of infection, to check that the strain remains susceptible.
The chief limitations of this study are that we were unable to test other strains exposed to cefiderocol in vivo and we did not compare our technique to the reference technique (PAP).
Cefiderocol remains a treatment of choice in cases of treatment failure in patients with CRAB infections. Determination of the type of carbapenemase produced by the infecting strain would help to guide therapeutic decisions, as in the case presented here. We further documented the presence/absence of adaptive resistance in clinical isolates, with the development of a new protocol that is more accessible to laboratories, but still requires a minimum of four days. However, we currently know little about the clinical impact of heteroresistance to cefiderocol, which has been a matter of debate in the scientific world since 2021 [4–10]. Future studies should investigate the determinants of cefiderocol resistance both in large series of OXA-type CRAB isolates tested with the reference method (PAP) and in large reviews/prospective studies of cefiderocol-treated infections [11]. Our study adds to knowledge of this phenomenon by linking the clinical course of the infection and microbiological investigations. We have three main conclusions: (i) for cefiderocol, EUCAST susceptibility testing is unable to detect adaptive resistance in Acinetobacter spp., (ii) knowledge of the carbapenemase produced by the CRAB isolate is useful, to guide the use of this drug, and (iii) the use of cefiderocol in monotherapy for OXA-23/24-producing CRAB should probably not be recommended, or should be considered only after an assessment of ARC. According to the WHO, research in this field is a matter of priority, to provide clinicians with guidance in the management of CRAB infections.