Acinetobacter baumannii has emerged as a major global threat, especially because of its exceptional ability to acquire resistance genes against all classes of antibiotics currently available. There have been high misidentification rates due to its phenotypic similarities with other Acinetobacter spp. However, differentiating between the species of the ACB complex, especially AB, is extremely important because of the poor outcomes and higher mortality rate in critically ill patients with invasive AB infections [17]. In our study, we found that second-generation commercial identification systems correctly identified only 55.6% of AB strains between 2001 and 2020. Before 2007, non-CC92 AB was predominant. However, after 2010, a complete replacement was observed from non-CC92 to CC92 genotypes at a single center. The AB CC92 isolates showed XDR characteristics, with only 5.8% of the isolates being susceptible to carbapenems. As carbapenem resistance increased, the proportion of patients treated with colistin also increased. The mortality rate decreased in 2018–2020 as majority of the infections were caused by colistin-susceptible ST784.

Automated phenotypic identification methods analyze reactions of the bacteria to different chemicals, creating an analytical biochemical profile, which then matches the profile to the “best fit” bacteria. First-generation automated bacterial identification systems included the analytical profile index (API) (bioMérieux, Craponne, France) and Vitek® system (bioMérieux, Craponne, France). Second-generation systems include MicroScan-Walkaway (Siemens, Munchen, Germany) and Vitek 2® (bioMérieux, Craponne, France), which are more accurate than first-generation systems [18]. However, in this study, second-generation commercial identification systems correctly identified only 55.6% (n = 47/93) of the AB isolated at a single center during 2001–2020, showing a misidentification rate of 44.4%. Similarly, a study evaluating the VITEK 2 System to identify AB showed that 68.0% of AB strains were correctly identified [19].

The correct identification of Acinetobacter spp. is crucial because of the poor prognosis of patients with AB infections, especially critically ill patients, compared with other non-baumannii Acinetobacter spp. infections [20,21,22,23]. In this study, there was a significant difference in 7-day mortality between AB and non-baumannii Acinetobacter spp., 40.0% vs. 4.2% (P < 0.001), and 30-day mortality (46.7% vs. 8.3%, P < 0.001). Differentiating between the species can allow clinicians to focus on early and aggressive interventions to enhance the survival and outcome of patients with invasive AB infections.

Data from Korean adults from the Korean Nationwide Surveillance of Antimicrobial Resistance (KONSAR) data in 2005 showed that the resistance rates of Acinetobacter spp. to imipenem and meropenem were 16% and 29%, respectively, and in 2011, the resistance rates increased to 64% and 63%, respectively [11, 12]. By 2015, the resistance rates were even higher, with the proportion of AB resistant to imipenem and meropenem being 85% and 84%, respectively [13]. Both AB and non-baumannii ACB complex Acinetobacter spp. were included in the adult data. However, in children with data including only invasive AB during similar time periods in this study, carbapenem resistance rates were similar during 2001–2009 and increased to > 90% after 2010. In the molecular epidemiology of both invasive and noninvasive ACB complexes isolated from children in Mexico, carbapenem resistance was 47% in 2017, and all XDR ACB complexes were AB, showing that carbapenem resistance is a problem in children as well as adults [24].

Pathogens identified as non-baumannii Acinetobacter spp. and AB non-CC92 isolates showed similar antibiotic susceptibility patterns, with susceptibility to cephalosporins, carbapenems, amikacin, and fluoroquinolones exceeding 85%, although the number of AB-non-CC92 isolates was limited. However, all AB CC92 isolates showed XDR characteristics, with complete resistance to 3rd/4th generation cephalosporins, piperacillin-tazobactam, and fluoroquinolones. Only 5.8% of isolates were susceptible to carbapenems. Susceptibility rates to colistin (40.4%) and amikacin (23.1%) were the highest, although both were below 50%.

Alarming levels of antibiotic resistance have been observed in AB worldwide. Resistance to carbapenems has been reported to reach 80.7% in Brazil, and only strains susceptible to polymyxins have been reported in Europe[25, 26]. In this study, since 2010, the increase in carbapenem resistance consequently led to increasing use of colistin in children with CRAB. However, between 2014 and 2017, a high mortality rate of up to 88% was observed in patients with invasive AB infections (Fig. 4). The high mortality can be attributed to clustered cases caused by both carbapenem and colistin-resistant pan-drug-resistant ST395. To date, there have been no previously published reports describing colistin resistance in ST395, and our study shows the need to monitor this pan-drug-resistant strain. During the consecutive period, 2018–2020, the mortality rate decreased to 19%. During this period, an outbreak of cases was caused by ST784, which was carbapenem-resistant and colistin-sensitive. However, factors associated with mortality are very complicated, as shown by the colistin resistance rate during 2001–2005, which was much higher than in 2018–2020, but with lower mortality rate. The high colistin resistance rate in 2001–2005 may have been driven by different resistance mechanisms than the colistin resistance after 2010, and further studies are needed to explain changes in mortality associated with antibiotics resistance.

The main mechanism underlying the poor outcome of children with invasive AB is their exceptional ability to acquire MDR-genes rapidly. As observed globally, there was a distinct change in the genotypes of AB causing invasive infections in children in this study. Before 2007, non-CC92 AB was predominant. However, after 2010, a complete replacement was observed from non-CC92 to CC92 genotypes at a single center. With genotype replacement, a significant increasing trend in carbapenem resistance and mortality was observed. This change has been reported globally, especially in China, where CC92 CRAB is increasing [27, 28]. Data on adults also show similar findings. A study including 19 different hospitals in South Korea reported a wide dissemination of ceftazidime resistant CC92 in South Korea [29]. During 2013–2017, wide distribution of CC92 and high prevalence of acquired carbapenemase genes among CRAB was reported in the USA [30]. Isolates collected during 2014–2015 from patients within a university hospital in southern Iran also reported the spread of closely related XDR genotypes of CC92 [31]. The epidemic dissemination of CC92 and near synchronous emergence worldwide in many countries are attributed to the successful and rapid acquisition of antimicrobial resistance genes [32, 33].

Furthermore, the pattern of circulating dominant STs within CC92 show two major patterns: in the case of ST1125, cases are observed throughout the entire study period. However, for majority of the STs, we see clustered cases or outbreaks during a certain timeframe within the study period. This latter pattern may be attributed to infection control measures applied to eradicate the ST causing clustered cases, hence elimination of the ST. However, due to the characteristics of AB, we see different STs continually emerging.

This study has several limitations. This was a single-center study; therefore, caution must be exercised in generalizing the data to other hospitals. However, our center has the largest pediatric ICU in the country, with 20 ICU beds, and is representative. Our data also show that the global trend of increase in CRAB CC92 strains applies even to hospital-acquired infections in children from a single center. Furthermore, this is the first study to investigate the 20-year long-term longitudinal molecular epidemiology of AB isolated from children with invasive infections at a single center and is therefore deemed to be valuable for monitoring and treating invasive AB infections in children. Further studies exploring the changes in resistance genes and virulence factors of these strains will provide insight into the changes in molecular epidemiology.