INTRODUCTION

Antibiotic resistance is a global threat to human health. Patients with infections by multidrug-resistant (MDR) Gram-negative bacilli (GNB) have an increased risk of poor outcome and mortality attributable to specific cause [1▪,2,3]. Over the past decades, the understanding of the mechanisms underlying antibiotic resistance has significantly increased [4,5]. Advances in molecular biology allowed the identification of complex resistance mechanisms able to confer a phenotype of resistance [6–8]. However, some phenomena, such as heteroresistance, remain difficult to be understood and their impact on clinical practice is still debated.

Heteroresistance is a phenotype in which a bacterial isolate contains subpopulations with reduced antibiotic susceptibility compared to the main population [9]. These populations pose a clinical concern, as they could become prevalent during or after antibiotic exposure. Heteroresistance was first described in the 1940 s when streptomycin resistance in type B Haemophilus influenzae emerged during exposure to this antibiotic leading to clinical failure [10]. Heteroresistance has been extensively studied for Gram-positive bacteria, in particular for heterogeneous vancomycin intermediate Staphylococcus aureus (hVISA). A recent meta-analysis showed a prevalence of 6% of hVISA among methicillin-resistant Staphylococcus aureus (MRSA) isolates [11,12]. Of note, there is heterogeneity in detection methods and controversial findings from available studies. Several clinical studies evaluated the association between hVISA and clinical failure, including persistent bacteraemia and treatment failure, but data on mortality are inconclusive [12].

Currently, heteroresistance among nonfermenting GNB drew the attention of scientific community because a debate about its potential role in explaining the clinical failure in some trials recently emerged [13▪▪,14]. Compared to hVISA, data about heteroresistance in nonfermenting GNB are limited and poorly evaluated.

In this review, we critically assess the published literature on heteroresistance in nonfermenting GNB. We underline that this is not a microbiological review, and we do not examine the specific mechanisms conferring heteroresistance. Conversely, we provide a clinical point of view and highlight the potential relationship between heteroresistance and clinical outcome of patients with infections by nonfermenting GNB. 

Box 1:

no caption available

DEFINITION OF HETERORESISTANCE

In general, heteroresistance is the presence of a heterogeneous population of bacteria with one subpopulation or several subpopulations that exhibit increased levels of antibiotic resistance compared with the main population [15]. However, this definition is incomplete lacking information of resistant subpopulations. The only type of heteroresistance with a clear consensus definition is the hVISA [defined as a S. aureus isolate susceptible to vancomycin (MIC 2 mg/l) but with minority populations (>10–6 cells] growing on vancomycin >2 mg/l by population analysis profile (PAP) investigation]. For other pathogens, especially for GNB, there is no global consensus on the definition of heteroresistance.

Figure 1 summarizes factors to be considered in the description of heteroresistance. The first factor is the clonality of resistant subpopulation. Both polyclonal and monoclonal heteroresistance may occur. In the first case (polyclonal heteroresistance), heteroresistance results from mixed infections or from rare resistant mutants that slowly increase in proportion during antibiotic treatment in a population of susceptible bacteria [9]. In the second case (monoclonal heteroresistance), heteroresistance may be generated from a single clone that differentiates into two populations (susceptible and resistant) at a high frequency in the absence of antibiotic pressure [9]. The second factor is the resistance level of the resistant subpopulations. Defining a minimal fold of increase in resistance for the subpopulation (defined as x-fold higher than the MIC of the main population) is important since it ensures that heteroresistance does not describe subpopulations with limited increases in MIC. Furthermore, heteroresistance may be classified as stable or unstable. The unstable heteroresistance occurs when the resistant subpopulations can revert to susceptibility in the absence of antibiotic pressure. Moreover, different mechanisms, including point mutations, insertion sequences and small deletions, may be involved in the develop of antibiotic heteroresistance and should be investigated.

FIGURE 1:

Factors to consider when defining antibiotic heteroresistance.

Finally, several detection methods, including the PAP test, the microdilution PAP test, the PAP–AUC, broth microdilution, E-test and disc diffusion, are used [16].

EVIDENCES FROM THE LITERATURE

To better summarize the available evidences about the relationship between heteroresistance and clinical outcome in infections by nonfermenting GNB, we performed a systematic review of the literature. A PubMed search (January 2003--July 2023) with the following searching terms (heteroresistance) AND (Gram negative OR Stenotrophomonas OR Acinetobacter OR Pseudomonas) was carried out. Narrative reviews were analysed to search for other relevant studies [15,17]. The flow diagram of the literature review is shown in Fig. 2. We identified a total of 148 studies. After screening by title/abstract and full text review, 42 studies evaluating heteroresistance in nonfermenting GNB were included. A summary of the main findings of the selected studies is reported in Table 1[13▪▪,14,18▪▪,19–56].

FIGURE 2:

Systematic review of studies evaluating the correlation between heteroresistance and clinical outcome in nonfermenting Gram-negative bacilli.

Table 1 - Summary of studies evaluating the correlation between heteroresistance and clinical outcome in nonfermenting Gram-negative bacilli (last 15 years) Ref. Pathogen Antibiotic Type of study Detection method % of HR Impact on clinical outcome Choby et al. [13▪▪] A. baumannii Cefiderocol Microbiological PAP 59% The authors hypothesized that clinical failure of CREDIBLE-CR is due to cefiderocol-HR among CRAB isolates. Choby et al. [14] A. baumannii Cefiderocol Microbiological PAP 20% The authors hypothesized that the better outcome of patients who received cefiderocol in the APEKS-NP trial compared to CREDIBLE-CR is due to the lower cefiderocol-HR in susceptible GNB (APEKS-NP). Stracquadanio et al. [18▪▪] A. baumannii Cefiderocol Microbiological PAP 80% -- Jo et al. [19] A. baumannii Colistin, tigecycline Microbiological PAP 100% -- Jo et al. [20] A. baumannii Tigecycline Microbiological PAP 56% -- Kon et al. [21] A. baumannii Colistin Clinical PAP 67.1% No significant increase in 14-day mortality or 14-day clinical failure, exception made for bacteraemia. Kon et al. [22] A. baumannii Colistin Microbiological PAP 100% -- Karakonstantis et al. [23] A. baumannii Colistin Systematic review and meta-analysis PAP 33% A single included study evaluating correlation between HR and clinical outcome Hong et al. [24] A. baumannii Colistin Microbiological PAP 50% -- Çağlan et al. [45] A. baumannii Colistin Microbiological PAP 21% Ezadi et al. [44] A. baumannii Colistin Microbiological PAP 21% Srinivas et al. [25] A. baumannii Colistin Clinical Mueller-Hinton agar plates 83% No difference in 30-day mortality between susceptible and HR strains. Choi et al. [26] A. baumannii Colistin A clinical case with multiple isolates PAP 84.2% of tested isolates from the same patient Clinical failure of the case Juhász et al. [43] A. baumannii
P. aeruginosa Colistin Microbiological Mueller–Hinton agar plates 20% A. baumannii
27% P. aeruginosa -- Rao et al. [27] A. baumannii Colistin Microbiological PAP 100% -- Rodriguez et al. [42] A. baumannii Colistin Microbiological PAP 95% -- Barin et al. [41] A. baumannii Colistin Microbiological PAP 90% -- Cai et al. [28] A. baumannii Colistin Systematic review Not reported 19–100% Combination therapy seems to be a strategic choice for HR isolates Rodriguez et al. [29] A. baumannii Colistin Microbiological PAP 86% -- Hung et al. [36] A. baumannii Cephalosporin Case series DD, Etest, PAP Previous exposure to cephalosporins may have selected HR-population Fernandez-Cuenca et al. [37] A. baumannii Carbapenem Microbiological DD IMI: 20%
MER: 25% Vidaillac et al. [40] A. baumannii
P. aeruginosa Colistin Microbiological PAP 100% Lee et al. [35] A. baumannii Imipenem Clinical DD, Etest 58% No difference in mortality, but higher exposure to IMI in controls Rodriguez et al. [30] A. baumannii Colistin Microbiological Time killing 43% . Rodriguez et al. [31] A. baumannii Colistin Microbiological PAP 46% Savini et al. [33] A.baumannii AMP/SUL Microbiological DD 25% Ikonomidis A et al. [34] A.baumannii Carbapenem Microbiologica PAP 100% of susceptible isolates Yau et al. [39] A. baumannii Colistin Microbiological PAP 23% Hawley et al. [32] A. baumannii Colistin Microbiological Mueller-Hinton agar 100% Li et al. [38] baumannii Colistin Microbiological PAP 94% Monogue et al. [56] P. aeruginosa C/T Microbiological PAP 23% Howard-Anderson et al. [47] P. aeruginosa Colistin Clinical PAP 6--26% No difference in 90-day mortality between susceptible and HR strains Li et al. [48] P. aeruginosa Levofloxacin Microbiological PAP 66.7% Jia et al. [49] P. aeruginosa Cefepime Clinical DD or Etest
then PAP 57% Treatment failure higher in infections by HR isolates. He et al. [50] P. aeruginosa Carbapenems Clinical Disk diffusion, PAP in 9 isolates IMI: 54%
MEM: 72%
IMI+MEM: 42% No difference of mortality between HR and susceptible strains. Hermes et al. [55] P. aeruginosa Colistin Microbiological PAP 4.2% Bergen et al. [51] P. aeruginosa Colistin Microbiological PAP 66.7% Pournaras S et al. [52] P. aeruginosa TZP Case report DD, PAP A single isolate before therapy start Favourable outcome reported in the first case of HR strain treated with TZP Ikonomidis A et al. [54] P. aeruginosa Carbapenems Microbiological PAP 100% Pournaras et al. [53] P. aeruginosa Carbapenems Microbiological DD, PAP 27.5% Walsh et al. [47] P. aeruginosa Fosfomycin Microbiological PAP 100%

AMP/SUL, ampicillin/sulbactam; C/T, ceftolozane/tazobactam; DD, disc diffusion; HR, heteroresistance; IMI, imipenem; MEM, meropenem; PAP, population analysis profile; TZP, piperacillin/tazobactam.

We highlight some important findings from the literature review:

(1) heteroresistance is commonly detected among nonfermenting GNB for several antibiotics, including polymyxins, β-lactams (carbapenems, piperacillin/tazobactam, cefiderocol), tigecycline, fosfomycin; (2) heteroresistance has been studied mainly for A. baumannii and P. aeruginosa, while no study was performed in S. maltophilia; (3) the majority of studies are microbiological experiments, while clinical studies evaluating the correlation between heteroresistance and clinical outcome of patients are limited (five for A. baumannii and four for P. aeruginosa) (4) there is heterogeneity in the detection methods.

These observations highlight that the knowledge of heteroresistance in nonfermenting GNB is less consolidated compared to Gram-positive bacteria. Data are heterogeneous and the phenomenon is so common for all tested antibiotics that it is difficult to evaluate its potential clinical role.

HETERORESISTANCE IN ACINETOBACTER BAUMANNII

Heteroresistance in A. baumannii is commonly detected for several antibiotics, including colistin, tygecicline, cefiderocol and ampicillin/sulbactam (Table 1).

Colistin-heteroresistance is the most common described type of heteroresistance in this pathogen. A meta-analysis including 15 studies showed an overall prevalence of colistin-heteroresistance among A. baumannii isolates of 33% [23]. Only one study on 24 patients evaluated the correlation of colistin-heteroresistance with clinical outcome and showed no differences in clinical cure and mortality between heteroresistant and homogeneously susceptible A. baumannii infections [23]. More recently, Kon et al. [21] investigated the prevalence of colistin-heteroresistance detected using the PAP and its evolution into full resistance among 173 clinical carbapenem-resistant A. baumannii (CRAB) isolates. Colistin-heteroresistance was detected in the 67.1% of CRAB isolates and the 80.2% of heteroresistant strains evolved into full resistance over time [21]. Of importance, in this study, colistin-heteroresistance was not significantly associated with 14-day mortality, with a trend for increased mortality observed only in patients with bacteraemia [21].

Heteroresistance in CRAB has been also detected for cefiderocol. Recently, some observational studies showed promising results in patients with CRAB infections treated with cefiderocol compared to colistin-containing regimens [57,58▪▪]. However, the CREDIBLE-CR trial showed higher clinical failure in patients treated with cefiderocol compared to those who received the best available therapy (BAT) [59]. Choby et al. [13▪▪,14] advanced the hypothesis that cefiderocol-heteroresistance may be the reason of the high clinical failure occurring in patients treated with cefiderocol. To support this hypothesis, they calculated the prevalence of cefiderocol-heteroresistance using the PAP test in CRAB isolates from the Georgia Emerging Infections Program (USA) [13▪▪]. They found that 59.3% of CRAB isolates showed cefiderocol-heteroresistance and postulated that heteroresistance might have contributed to cefiderocol treatment failure in the CREDIBLE-CR study [13▪▪]. Moreover, to explain the discrepancy of mortality rates between the CREDIBLE-CR trial and the APEKS-NP trial (including carbapenem susceptible GNB) [60], in a further experiment, they calculated the prevalence of cefiderocol-heteroresistance among carbapenem susceptible GNB from the same Georgia cohort [14]. They observed lower rates of cefiderocol-heteroresistance in carbapebem-susceptible GNB compared to those detected in the previous study (11% in fully susceptible and 44% in cephalosporin-resistant isolates), postulating that clinical failure of cefiderocol depends on the different rates of heteroresistance [14].

The debate is still open for several reasons [61]. First, data of cefiderocol-heteroresistance in CRAB come from a cohort different from CREDIBLE-CR (Georgia surveillance). Second, despite the high prevalence of colistin-heteroresistance among CRAB, heteroresistance rate and its correlation with clinical outcome in the BAT group have not been considered. Finally, there are other relevant factors that may have contributed to the increased mortality observed in the cefiderocol-treated patients of the trial, who had more frequently septic shock and were more commonly cared for in intensive care unit compared to BAT. A recent microbiological study elucidated important characteristics of cefiderocol-heteroresistant in CRAB isolates [18▪▪]: first, high rates of cefiderocol-heteroresistance after exposure to this drug was detected (heteroresistance detected in eight out of 10 isolates using the PAP method); second, small colonies immediately turned back to their original shape once the drug was removed, suggesting that cefiderocol-heteroresistance is unstable; third, since the strains harboured multiple classes of b-lactamase resistance genes, the addition of β-lactamase inhibitors, such as ceftazidime/avibactam, to cefiderocol may restore its activity [18▪▪]. Heteroresistance has been also detected among A. baumannii isolates for other antibiotics, including tigecycline (up to 56%) [18▪▪] and ampicillin/sulbactam (25% in a single study) [33].

HETERORESISTANCE IN PSEUDOMONAS AERUGINOSA

In P. aeruginosa, heteroresistance has been reported in different studies [47–56], and its role in treatment failure was reported in a large retrospective cohort, where imipenem and meropenem heteroresistance was detected in 54 and 73% of the isolates, respectively [50]. Heteroresistance rates for colistin and levofloxacin are higher than 60% [46–48], while a single clinical case of a patients with infection by P. aeruginosa with heteroresistance to piperacillin/tazobactam that no lead to clinical failure has been reported [52]. Recently, ceftolozane/tazobactam-heteroresistance has been reported, with a prevalence of 23% [56]. Available studies did not demonstrate that heteroresistance in P. aeruginosa may increase mortality.

EXPERT OPINION ON CLINICAL IMPLICATIONS OF HETERORESISTANCE IN NONFERMENTING GRAM-NEGATIVE BACILLI

Current available data about the role of heteroresistance among nonfermenting GNB are limited and do not clearly demonstrate its potential clinical role. Anyway, the literature review suggests the following considerations: first, unlike data from hVISA, heteroresistance is widely detected among nonfermenting GNB isolates. In some studies, almost all isolates were found to be heteroresistant even in isolates without prior antibiotic exposure. This first observation highlights that heteroresistance may be an inherent characteristic of these strains. With this in mind, it is difficult to evaluate the impact of this common phenomenon on the clinical practice.

Second, lack of a standard definition of heteroresistance may lead to misidentification of homogeneous strains as heteroresistant, hindering proper assessment of its clinical relevance. Moreover, the best available method to detect in-vitro heteroresistance is the PAP test, but it is a highly time and workforce-consuming method, needs for skilled personnel and performed by reference laboratories. While a standardized detection test is available for hVISA, for nonfermenting GNB detection methods should be further validated.

Third, it is not known whether the presence of in-vitro heteroresistance may lead to in-vivo selection of resistant isolates. Resistant subpopulations with significant fitness cost may be unstable and not sustained over time. Emergence of cefiderocol resistance among CRAB is of clinical concern [62▪,63]. It is not known if underlying mechanisms are shared between heteroresistant subpopulations and fully resistant isolates. Further studies are needed to describe molecular mechanisms of cefiderocol-heteroresistance among CRAB isolates and to evaluate whether it may be predictive of emergence of resistance during cefiderocol exposure.

Finally, no data demonstrated a relationship between the in-vitro heteroresistance and in-vivo clinical outcome. However, heteroresistance may have clinical implications. It may contribute to treatment failure in particular clinical situations, such as high-inoculum infections or immunocompromised patients. In these cases, antibiotic combination regimens instead of monotherapy may be more appropriate, especially during the first phases of infection.

The presence of heteroresistance among nonfermenting GNB should be translated into clinical practice to prefer combination therapy over monotherapy to prevent the selection of resistant subpopulations during treatment. The use of combination therapy is largely debated in the scientific community. With concern to CRAB, combination regimens are usually used and also suggested by current guidelines in severe infections [64–66]. In an observational study from our group, among patients treated with CRAB infections treated with cefiderocol who experienced microbiological failure, 50% developed resistance to the siderophore antibiotic [58▪▪]. The rate of microbiological failure was significantly higher in patients who received monotherapy compared to those who received combination therapy [58▪▪]. Thus, awaiting for further studies to better understand the microbiological and clinical role of heteroresistance, caution should be taken for monotherapy among CRAB isolates. Conversely, combination therapy should be preferred when both a colistin-based or cefiderocol-based therapy is chosen. Combination of colistin or cefiderocol with other antibiotics, such as ampicillin/sulbactam, should be preferred for CRAB infections.

In patients with invasive infections caused by P. aeruginosa, combination therapy should not be the routine choice but may be considered on a case-by case basis, especially in high-inoculum or in the first phases of infection [65].

CONCLUSION

Unlike Gram-positive bacteria, heteroresistance is a common in-vitro phenomenon in nonfermenting GNB. It has been studied mainly for A. baumannii and P. aeruginosa and appears to be frequent for several tested antibiotics, including colistin, tygecicline, cefiderocol, carbapenems. The PAP is the gold standard to detect heteroresistance, but its use has been validated only for hVISA. It is not known if the presence of in-vitro heteroresistance may have a clinical impact on clinical outcome in vivo. Studies evaluating this potential correlation are limited and inconclusive.

Since heteroresistance is very common among nonfermenting GNB, it should be considered an inherent characteristic of these strains and cannot be directly predictive of clinical failure. Further studies are needed to better elucidate this point. Unmet needs in this field are represented by the understanding of molecular mechanisms underlying heteroresistance and its role in the potential development of treatment emergent resistance. Of importance, the presence of heteroresistance among nonfermenting GNB should lead to prefer combination regimens instead of monotherapy in the clinical practice, especially in high-inoculum infections.

Acknowledgements

None.

Financial support and sponsorship

None.

Conflicts of interest

G.T. received honoraria by Shionogi for educational meetings. M.F. received unconditional grants from MSD and grants/or speaker honoraria from Angelini, Shionogi, Pfizer, Menarini, Gilead, TermoFisher and Nordic Pharma. The remaining authors have no conflicts of interest.

REFERENCES AND RECOMMENDED READING

Papers of particular interest, published within the annual period of review, have been highlighted as:

▪ of special interest

▪▪ of outstanding interest

REFERENCES 1▪. Falcone M, Tiseo G, Carbonara S, et al. Mortality attributable to bloodstream infections caused by different carbapenem-resistant Gram-Negative Bacilli: results from a nationwide study in Italy (ALARICO Network). Clin Infect Dis 2023; 76:2059–2069. 2. Falcone M, Tiseo G, Durante-Mangoni E, et al. Risk Factors and outcomes of endocarditis due to non-HACEK Gram-negative bacilli: data from the prospective multicenter Italian Endocarditis Study Cohort. Antimicrob Agents Chemother 2018; 62:e02208–e2217. 3. Falcone M, Tiseo G, Dentali F, et al. Predicting resistant etiology in hospitalized patients with blood cultures positive for Gram-negative bacilli. Eur J Intern Med 2018; 53:21–28. 4. Tiseo G, Falcone M, Leonildi A, et al. Meropenem-vaborbactam as salvage therapy for ceftazidime-avibactam-, cefiderocol-resistant ST-512 Klebsiella pneumoniae-producing KPC-31, a D179Y variant of KPC-3. Open Forum Infect Dis 2021; 8:ofab141. 5. Falcone M, Giordano C, Barnini S, et al. Extremely drug-resistant NDM-9-producing ST147 Klebsiella pneumoniae causing infections in Italy, May 2020. Euro Surveill 2020; 25:2001779. 6. Kumarasamy KK, Toleman MA, Walsh TR, et al. Emergence of a new antibiotic resistance mechanism in India, Pakistan, and the UK: a molecular, biological, and epidemiological study. Lancet Infect Dis 2010; 10:597–602. 7. Falcone M, Tiseo G, Galfo V, et al. Bloodstream infections in patients with rectal colonization by Klebsiella pneumoniae producing different type of carbapenemases: a prospective, cohort study (CHIMERA study). Clin Microbiol Infect 2022; 28:298.e1–298.e7. 8. Falcone M, Tiseo G, Antonelli A, et al. Clinical features and outcomes of bloodstream infections caused by New Delhi metallo-β-lactamase-producing Enterobacterales during a regional outbreak. Open Forum Infect Dis 2020; 7:ofaa011. 9. Andersson DI, Nicoloff H, Hjort K. Mechanisms and clinical relevance of bacterial heteroresistance. Nat Rev Microbiol 2019; 17:479–496.