The Centers for Disease Control and Prevention (CDC) defines CRE as Enterobacteriaceae that are resistant to carbapenem antibiotics. In the United States, approximately 13,000 infections caused by CRE have been reported in hospitalized patients, resulting in an estimated 1,100 deaths (20). Patients who require medical devices such as ventilators, urinary catheters, or intravenous catheters, those who are on prolonged antibiotic treatment, and individuals with weakened immune systems are at a relatively high risk of contracting CRE infections (21). Hence, it is crucial to exercise caution in implementing therapeutic measures to avoid unnecessary invasive procedures for patients who usually have underlying medical conditions and those who require ICU-level interventions. Based on the resistance mechanism, CRE can be categorized into carbapenemase-producing enterobacteria (CPE) and non-carbapenemase-producing enterobacteria (non-CPE). CPE comprises carbapenemase-resistant enzymes such as K. pneumoniae carbapenemase (KPC) in class A, benzoxacillin carbapenemase/oxacillinase (OXA) in class D, and MBLs belonging to class B, including imipenemase metallo-β-lactamase (IMP), New Delhi metallo-β-lactamase (NDM), and Verona integrase-encoded metallo-β-lactamase (VIM) (11). The bla_KPC gene is the most prevalent gene in Ambler class A, and its production plays a significant role in the carbapenem-resistant K. pneumoniae (CRKp). On the other hand, bla_OXA-48 is a more common gene responsible for resistance to carbapenem antibiotics in Escherichia coli (22). Single CRE isolate can possess multiple carbapenemase-encoding genes. For instance, in Egypt, where NDM and OXA-48-like enzymes are widespread, polymerase chain reaction results demonstrated that about 90% of Enterobacteriaceae isolates harbored one or more carbapenemase-encoding genes, with bla_NDM-1 being the most prevalent genotype, followed by bla_OXA-48 (23).

Acinetobacter baumannii (A. baumannii) is the most frequently isolated pathogen in ICUs (24), leading to various infections such as pneumonia, skin and soft tissue infections, and bloodstream infections (BSIs) (25, 26). The emergence of CRAB poses a significant challenge for treatment and has intensified the prevalence of hospital-acquired infections, thus becoming a major threat to global public health (11). The mechanisms of carbapenem resistance in A. baumannii involve various factors such as increased efflux pumps, decreased expression or inactivation of pore proteins, modifications of penicillin-binding proteins, and production of several types of β-lactamases (27, 28). The most common mechanism observed in CRAB is the production of carbapenemases, and the genes encoding the acquired carbapenemases play a key role. Among carbapenemases, the OXA enzymes are the most frequently reported in A. baumannii, such as OXA-23, OXA-24, OXA-40, OXA-51, OXA-58, and OXA-143 (29, 30). Additionally, metastable MBLs, including VIM, IMP, and NDM enzymes, have also been linked with drug resistance phenotypes in A. baumannii (11). It is essential to note that although KPC enzymes have primarily been detected in K. pneumoniae, variants of bla_KPC, such as bla_KPC-2 and bla_KPC-3, have been reported in A. baumannii in a Brazilian hospital (31). The acquisition of bla_KPC might be associated with A. baumannii’s resistance to carbapenems.

Pseudomonas aeruginosa (P. aeruginosa) is a Gram-negative bacterium commonly found in moist environments, such as washing tanks, aerators, respirators, and other equipment, as well as solutions exposed in hospital environments (32). It is a significant cause of healthcare-related problems, leading to urinary, respiratory, and BSIs in long-stay hospitalized patients (11, 33). These infections can be fatal in critically ill and immunocompromised patients in ICUs, and they may be further exacerbated by AMR (34). Infections caused by CRPA result in longer hospitalization periods and higher mortality rates compared to infections caused by carbapenem antibiotic-sensitive strains (35, 36). The development of CRPA involves the interaction of several complex resistance mechanisms. Firstly, the upregulation of efflux pumps (e.g., MexAB-oprM) allows for increased drug efflux, leading to resistance against most β-lactams (37). Additionally, the loss of OprD outer membrane proteins, which normally prevent the entry of antibiotics, coupled with the overproduction of Ambler C-like enzymes, can result in the near-exhaustion of P. aeruginosa’s resistance to β-lactams (38). Resistance to carbapenem antibiotics through carbapenemase production is a less common mechanism (39). Out of 28 CRPA strains isolated in the ICU, only three strains produced KPC (40). However, carbapenemase production as a resistance mechanism appears to be increasingly common, with bla_VIM in MBLs being the most commonly detected gene, typically encoded on plasmids that are highly capable of dissemination (38). In instances where CRPA lacks carbapenemases, resistance is typically due to the absence of OprD or the overexpression of efflux pumps.

Avibactam binds reversibly to β-lactamases and exhibits activity against carbapenemases, thereby restoring the inhibitory activity of ceftazidime against the majority of CRE and CRPA. Ceftazidime/avibactam generally demonstrates high efficacy against organisms producing KPC, although resistance has been observed in isolates producing KPC-2 and KPC-3, which may be attributable to reduced porin expression (41). Combination therapy could reduce mortality in BSIs caused by KPC-producing K. pneumoniae (42). However, the use of ceftazidime/avibactam in combination with other antimicrobial agents for the treatment of CRE and CRPA infections did not exhibit significant advantages in terms of survival and cure rates (43, 44). Ceftazidime/avibactam alone demonstrates superior effectiveness in patients with OXA-48-producing CRE infections compared to treatment with colistin, tigecycline, and meropenem (45, 46). Therefore, prioritizing ceftazidime/avibactam for the treatment of KPC-producing or OXA-48-producing CRE, as well as CRPA, may improve survival rates among patients in the ICU and reduce the risk of renal injury, as opposed to selecting alternative drugs or multidrug combinations. Moreover, since avibactam does not inhibit MBLs (NDM, VIM, and IMP), combining it with aztreonam, a drug stable against metallo-β-lactamases, may be a potential therapeutic strategy for treating CR-GNB infections belonging to class B. The combination of ceftazidime/avibactam and aztreonam exhibits good in vitro activity against Enterobacteriaceae producing metallo-β-lactamases, with favorable in vitro effectiveness (47).

Vaborbactam is a β-lactamase inhibitor that primarily targets KPC carbapenemases but not MBLs, as well as class D β-lactamases (48). On the other hand, meropenem effectively treats Gram-negative bacilli, such as K. pneumoniae, Enterobacter spp., and P. aeruginosa. Together, meropenem/vaborbactam is a novel combination that exhibits strong and specific activity against KPC-producing CRE. While Vaborbactam also possesses the capacity to inhibit ESBLs and AmpC β-lactamases, its supplementary activity is not necessary, as meropenem alone effectively stabilizes these β-lactamases. Though multi-agent treatments may benefit high-risk patients, mono-therapy may be enough for other patients. For instance, meropenem/vaborbactam alone showed higher cure rates and lower patient mortality and nephrotoxicity in individuals with predominantly bacteremic CRE infections compared to other drug combinations (49). Therefore, along with considering the type of carbapenemase, a successful treatment of CR-GNB infections also necessitates consideration of different infection types, the severity of the infection, susceptibility of the causative organism, and the patient’s general health condition. Thus, meropenem/vaborbactam is another viable option for KPC-producing CRE infections.

Relebactam is a type of β-lactamase inhibitor, structurally similar to avibactam, that can inhibit common class A carbapenemases (e.g., KPC) and class C cephalosporinases (e.g., AmpC). In vitro, relebactam has been shown to reverse resistance to imipenem in KPC-producing P. aeruginosa but has no potentiation effect in isolates with class B or D carbapenemase activity (50). In an in vivo C. elegans model, imipenem/relebactam was found to be a significant treatment for KPC-producing K. pneumoniae infections (51). Furthermore, relebactam in combination with imipenem/cilastatin inhibited AmpC, thus restoring the susceptibility of P. aeruginosa to imipenem. The combination demonstrated better efficacy with lower mortality and nephrotoxicity for the treatment of patients infected with CR-GNB (52). It is worth noting that this study included high-risk patients with poor outcomes, and the combination of relebactam and IMI is a potential therapeutic option for ICU patients infected with CR-GNB. Against blaKPC-containing P. aeruginosa, Imipenem/relebactam was shown in vitro to have a higher inhibitory activity than meropenem/vaborbactam but lower than ceftazidime/avibactam (53). Therefore, rapid diagnosis of the carbapenemase genotype of CRE or CRPA is significant for β-lactam/β-lactamase inhibitor combinations (BL/BLI) selections in clinical settings.

The recently approved BL/BLIs expand the therapeutic options available for KPC-producing and OXA-48-producing Enterobacteriaceae. Cefiderocol, the first cephalosporin containing an iron-based carrier, has gained approval for the treatment of carbapenem-resistant non-fermenting bacteria, including P. aeruginosa and A. baumannii (54). Current studies have demonstrated significant in vitro activity and effectiveness of cefodilol against CR-GNB (55). Cefodilol exhibits inherent stability against a wide range of carbapenemases, including class A, B, and D, as well as class C cephalosporinase hydrolases. Patients infected with KPC, NDM, VIM, IMP, and OXA-48 harboring CRE experiencing BSI or urinary tract infections can potentially benefit from cefodilol therapy (56). However, it is important to acknowledge that resistance may arise when cefiderocol is employed in the treatment of CRE. Instances of cefiderocol resistance in these isolates can be attributed to factors such as the clinical environment, in vitro exposure to cefiderocol, or resistance to other β-lactam antibiotics (e.g., ceftazidime or cefepime) prior to treatment. The application of cefiderocol in these cases carries a risk of mutation resulting in the development of specific mutations, such as NDM-5 (57), KPC-41, KPC-50 (58), and OXA-427 (59).

A study by Falcone et al. (60) conducted in the ICU featured 10 patients with A. baumannii BSI and ventilator-acquired pneumonia. These individuals had previous treatment failures with antibiotics, including colistin, and had developed renal and hepatic injury. Clinical success and survival rates at 30 days were 70 and 90%, respectively, with cefiderocol treatment. Cefiderocol monotherapy for critically ill patients was revealed to result in a lower infection recurrence rate and higher clinical success compared to combination therapy using drugs like colistin (61). However, all-cause mortality was higher with cefiderocol monotherapy. This trend may be linked to the heightened risk of infection recurrence or death in critically ill patients, who commonly experience trauma-induced immune compromise, prolonged hospitalization, invasive procedures, and colonization of the skin by multidrug-resistant organisms. Consequently, in addition to the timely and accurate selection of appropriate drug therapy, implementing specific preventive and control measures against CR-GNB infections in the ICU setting is paramount.

The newly approved BL-BLIs have emerged as the primary treatment options for CRE and CRPA infections. However, the treatment landscape for CRAB infections is becoming increasingly limited. Although BL-BLI therapy is recommended for urinary tract infections due to the high concentration of polymyxins in the urinary tract, it is still considered an alternative therapy for CRAB infections (62). According to the 2023 guidelines by the Infectious Diseases Society of America (IDSA) for the management of resistant Gram-negative bacteria infections, high-dose ampicillin-sulbactam in combination with other agents, including Polymyxin B, is recommended for the treatment of moderate-to-severe CRAB infections (63). Polymyxin B is specifically indicated for the treatment of severe infections, such as BSIs. However, the efficacy and safety of Polymyxin B as a monotherapy are not well-established. Thus, it is generally advised to administer Polymyxin B in combination with at least one other antimicrobial agent from a different class. Colistin, another approved drug from the polymyxin class of antibiotics, is considered a last-resort treatment for A. baumannii infections (64).

However, the emergence of polymyxin-resistant strains has been well-documented, potentially due to colistin’s prodrug nature and the prolonged presence of its active form in the body, which can predispose to resistance (65). Typically, colistin is recommended in combination with other agents for the treatment of CRE with CRPA infections. However, a clinical observational study (66) found that the difference in outcomes between patients treated with colistin sulfate alone versus in combination with other antibiotics was not statistically significant. The use of low-dose polymyxins in the treatment of multidrug-resistant A. baumannii infections may elevate the risk of mortality (66). A trial in 2018 demonstrated that colistin combined with meropenem treatment did not yield improved outcomes for severe infections caused by CR-GNB (67). Additionally, colistin therapy, especially when employed as part of combination therapy for patients with CR-GNB infections, may result in unfavorable clinical outcomes and potentially increase the risk of kidney injury in patients (68, 69). Combination therapy could heighten the probability of adverse effects, escalate the cost of antimicrobial therapy, and contribute to the development of antimicrobial resistance. Further clinical trials are imperative to establish the efficacy and safety of colistin as a complementary or alternative treatment for severe CR-GNB infections, particularly in cases where BL-BLI treatment proves ineffective.

Tigecycline is a novel intravenous antibiotic with broad-spectrum activity and is derived from minocycline. It has traditionally been considered the preferred treatment for infections caused by CRE. However, the latest guidelines from the IDSA recommend β-lactams as the primary option for treating CRE infections, with tigecycline as an alternate option if necessary (63). Combining tigecycline with colistin, carbapenems, or aminoglycosides is the most commonly used regimen for treating CRE infections. Studies comparing these combinations found that tigecycline-colistin was most effective against Klebsiella, while imipenem-colistin was best against Escherichia coli (70). Tigecycline combined with amikacin and colistin, or minocycline with cefoperazone-sulbactam, showed synergistic inhibitory activity against CRAB (71, 72). OXA-24-producing strains are more sensitive to tigecycline-amikacin and OXA-23-producing strains are more sensitive to tigecycline-mucin use [1]. Minocycline and tigecycline have lower nephrotoxicity compared to mucins versus aminoglycosides and can be used in combination with other drugs as another treatment option for CRAB (72). Notably, tigecycline-based regimens with high-doses (200 mg loading and 100 mg maintenance) showed lower mortality rates in ICU patients than standard doses (100 mg loading and 50 mg maintenance), and combination therapy with tigecycline was more effective than monotherapy (73). Consequent to exposure to tigecycline, resistance was induced in CRKp but tigecycline-resistant strains exhibited greater susceptibility to other drugs, including aminoglycosides, carbapenems, and cephalosporins (74). Sequential combination therapy with tigecycline and aminoglycosides may be a more effective approach to treating CRE.

Aminoglycoside antibiotics possess strong bactericidal properties and remain effective in treating MDR-GNB. However, their application is somewhat limited due to the side effect of nephrotoxicity. Generally, aminoglycosides are not the primary treatment option for severe infections. However, they can still be considered as a therapeutic alternative for combating CR-GNB when other options are unavailable. This is usually done in combination with other drugs such as β-lactams (75). For instance, studies have shown that the combination of imipenem and amikacin has a synergistic effect on CR-GNB both in vivo and in vitro (76–78). Amikacin exhibits lower resistance than gentamicin in most CRE strains (79). Many studies have supported the use of aminoglycosides in the treatment of CRE infections in critically ill patients before the introduction of novel BL-BLIs (80). Furthermore, gentamicin has demonstrated the potential to reduce mortality in K. pneumoniae sepsis caused by class A β-lactamase-producing enzymes, including KPC-3, SHV-11, and TEM-1 (81). A recent case report highlighted successful treatment of a patient with a CRKp intracranial infection after craniotomy using intrathecal injection of gentamicin and intravenous injection of amikacin, which displayed gentamicin susceptibility (82).

Plazomicin, a next-generation aminoglycoside antibiotic, has demonstrated a lower minimum inhibitory concentration compared to other aminoglycosides, making it a potential treatment option for infections caused by carbapenemase-producing, NDM-producing CRE (83). In a multicenter, randomized, open-label phase III trial that compared plazomicin with colistin (both in combination with imipenem) for the treatment of severe infections in CRE, plazomicin proved to be effective with a relatively low mortality and complication rate (84). Due to the nephrotoxicity associated with aminoglycoside antibiotics, they are generally not used in combination with colistin. To ensure optimal efficacy and minimal toxicity, appropriate dosage, administration, and therapeutic drug monitoring of the patient are essential when using aminoglycosides.

Hand hygiene: Multi-drug resistant organisms (MDROs) have exhibited the ability to persist in hospital environments, such as floors, walls, beds, doorknobs, bedside tables, and equipment (86). Barnes et al. (87) developed a patient–patient transmission model within the ICU and compared the effects of hand hygiene and environmental cleanliness on MDRO acquisition rates; findings suggest that universal decolonization methods could eliminate colonization of MDRO Gram-positive bacteria. For example, patients in the ICU receiving mupirocin nasal injection have lower rates of MRSA BSIs compared to those undergoing chlorhexidine bathing (88). Extensively resistant MDROs, such as CRE, have shown poor response to chlorhexidine treatment, and current clinical evidence does not support the removal of patient colonization (89, 90). A meta-analysis indicated that ICU bathing with chlorhexidine significantly reduces A. baumannii colonization (91). Compliance with hand hygiene is widely considered as the foundation for preventing MDRO spread in ICUs. However, in hospitals with low compliance rates, proactive detection of CR-GNB has substantial benefits for patients when implemented with increased environmental cleanliness. Nevertheless, in high hand hygiene compliance environments, contact precautions and screening for CR-GNB colonization contribute little to preventing MDRO spread, especially for CR-GNB.

Antimicrobial Stewardship: Antimicrobial Stewardship (AMS) is a significant measure of importance as defined by the IDSA. It entails implementing coordinated interventions aimed at enhancing and evaluating proper utilization of antimicrobials. This is accomplished through facilitating optimal selection of antimicrobial regimens, determining appropriate dosage, therapy duration, and administration route. For patients in the ICU, the potential negative consequences of antimicrobial overuse are considered less perilous compared to the inadequate employment of restraints. Reports indicate a substantial proportion of ICU patients receiving excessive antimicrobial therapy, which includes treatment involving antimicrobials for suspected infections, utilization of overly broad-spectrum antibiotics, delayed initiation of timely antibiotic de-escalation and optimization, and prolonged duration of therapy (92, 93). The implementation of Antibiotic Stewardship Programs (ASP) within the ICU setting can potentially reduce the misuse of antimicrobials, shorten hospital stays, minimize costs, and decrease the emergence of drug resistance (94). Also, the study conducted by Khdour et al. (95) highlights the importance of establishing a well-structured antimicrobial stewardship team in the context of AMS. They found that timely feedback and prospective audits from the antibiotic stewardship team, within 48–72 h of antibiotic administration for ICU patients, had a positive impact on patient outcomes. Calcitoninogen as a biomarker in the ICU has been shown to reduce the use of antibiotics and mortality rates to some extent (96). However, further investigation is needed to determine the efficacy of calcitonin as a treatment indicator, and the cost of frequent testing must be balanced with potential savings from shorter antibiotic therapy. To address the growing issue of carbapenem resistance, experts emphasize the importance of implementing clear strategies to guide the appropriate use of carbapenem antibiotics (97).

Rapid Diagnostic Tests: Standard microbial identification techniques typically take 48–72 h, while optimizing antibiotic therapy within the first 6–12 h of infection is critical for treating life-threatening infections. Rapid diagnostic tests (RDTs) provide assistance to ASP by contributing to timely and effective antimicrobial therapy, potentially reducing mortality, hospitalization, and costs, as well as improving antimicrobial use and clinical and economic outcomes. Recently developed RDTs are able to provide identification results within 3 h of collection and 2.5 h after Gram staining (98). The RDTs provide an opportunity to rapidly optimize antimicrobial therapy, but have been shown to be combined with ASP to maximize translation into improved patient outcomes (99). Studies have identified genotyping and phenotyping of Escherichia coli, Klebsiella, etc. based on RDTs to predict susceptibility to β-lactams (ceftazidime, piperacillin-tazobactam, imipenem, and meropenem). RDTs can support downgrading decisions for the treatment of GNB infections (99).

Screening and prophylactic isolation: Patients in the ICU are particularly vulnerable to colonization or infection with MDRO either upon admission or during their hospital stay due to various risk factors. To reduce the spread of MDRO, it is crucial to implement proactive screening or isolate patients with high-risk factors (100). Although proactive testing methods differ among hospitals in different regions, they usually involve obtaining fecal/rectal swabs from patients upon admission or at regular intervals (weekly or bi-weekly). This practice applies to all patients or those at high risk (e.g., ICU patients, those with a history of previous colonization/infection), with a focus on identifying CRE. Results of a study revealed a high incidence of CRKp colonization and a likelihood of eventual CRKp infection in patients who carried Klebsiella pneumoniae (including CRKp or carbapenem antibiotic-susceptible Klebsiella pneumoniae) upon ICU admission (101). Proactive screening in high-risk units for CRE colonization or infection has also shown that CRE-positive patients, both neonatal and non-neonatal, exhibit different genotypes of carbapenemases. Notably, over 90% of CRE-positive neonates carry NDM. Isolating and placing these patients appropriately may help reduce the risk of CRE infection (102).

Additionally, the implementation of proactive testing and isolation strategies has shown a decrease in infections caused by CRAB and CRPA at a broader scale (103). Hospitals with limited isolation facilities have commonly resorted to a contact precautions approach in confining ICU patients to their own beds, similar to horizontal measures. Notably, no transmission of resistant organisms, such as Klebsiella pneumoniae, was detected in these cases (89). Implementing universal contact precautions, regardless of the specific pathogen, may also impede the spread of CR-GNB within the ICU.

The emergence and spread of CR-GNB are influenced by various factors. Identifying high-risk factors to determine ideal target populations for proactive testing or prophylactic contact helps optimize the allocation of limited resources. Depending on the purpose of the prediction model, different target populations can be selected for retrospective or prospective studies. The CR-GNB infection prediction model and the early warning model in Table 2 were designed to identify patients infected with or carrying CR-GNB, respectively. In order to prevent the spread of CR-GNB before culture results are available, modeling techniques have been employed to assist in the pre-isolation of potential carriers of CR-GNB or patients who are at a high risk of infection upon admission. There is also a type of predictive modeling that predicts infection at a particular site. BSI is a severe infection characterized by positive blood cultures in patients displaying symptoms of systemic infection. BSI often leads to unfavorable outcomes for patients in the ICU, including longer hospital stays and higher mortality rates (109, 110). While blood cultures serve as the gold standard and primary tool for diagnosing pathogens causing BSI, they are susceptible to delays in initiating effective treatment due to the time required (111). In addition to performing timely blood cultures or rapid diagnostic tests when BSI is suspected, several studies have explored the use of predictive modeling to construct early warning models for BSI. Several studies, as depicted in Table 3, have developed early warning models to identify BSI in vulnerable populations, such as children, the older adults, and individuals with immunodeficiencies. These models rely on risk factors or biomarkers to target high-risk populations and implement prophylactic measures, thereby reducing the occurrence of BSI and the associated mortality risk. While most of these models have demonstrated reliable predictive performance, unfortunately, only a limited number of studies have conducted validation in diverse healthcare settings. Consequently, the geographical applicability of these models may be constrained due to this lack of validation across multiple centers.