The isolates evaluated for K. pneumoniae were identified by the Microbiology Department of the Clinical Laboratory at Hospital Israelita Albert Einstein between January 2018 and January 2021. The selected isolates were obtained from patients admitted to the ICU or semi-ICU. Isolates from blood, bronchoalveolar lavage, and tracheal secretions were considered infection-associated isolates for the purpose of this study, and isolates from rectal swabs as part of the institutional multidrug-resistant bacteria surveillance program were carriage isolates. When an isolate of infection or colonization was reported from the same patient with the same collection date, isolation site, and susceptibility profile, only the first report of the isolate was retained for further analysis.

Species confirmation of K. pneumoniae was performed using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) (Bruker Daltonics, Billerica, MA, USA). Antimicrobial profile evaluation was conducted using the disk diffusion method for imipenem and meropenem in surveillance cultures (colonization samples) and the automated VITEK 2 system (bioMérieux, France) for infection samples. The VITEK 2 system includes the following antimicrobials: ceftazidime, cefepime, amikacin, gentamicin, ciprofloxacin, imipenem, and meropenem, with carbapenem resistance confirmed using gradient diffusion methods. The antimicrobial susceptibility results were interpreted according to the Brazilian Committee on Antimicrobial Susceptibility Testing/European Committee on Antimicrobial Susceptibility Testing guidelines for each year (http://brcast.org.br/). The proportion of resistant isolates was calculated by dividing the number of resistant K. pneumoniae isolates by the total number of isolates tested against the corresponding antibiotics and multiplying by 100.^{( 20 )} The endemic level of K. pneumoniae isolates was calculated per 10,000 patient-days, according to the method described by Arantes et al.^{( 21 )} The presence of the bla _KPC gene in isolates showing reduced susceptibility to carbapenems was determined by real-time polymerase chain reaction (PCR).^{( 22 )}

To characterize the population of K. pneumoniae carrying bla _KPC (KpKPC) circulating in the hospital while avoiding community-acquired isolates, molecular analyses were restricted to KpKPC from colonized patients or those with infections admitted to the ICU or semi-ICU during the periods of highest KpKPC prevalence and adjacent periods between January 2018 and January 2021. Furthermore, bacterial isolates for molecular analyses were selected under two conditions: availability of isolates in the isolate bank of the Clinical Laboratory and bacterial viability. Only one isolate per patient was included in the analysis, corresponding to the first colonization and/or infection sample positive for KpKPC.

Genetic relatedness was established using PFGE as described previously.^{( 23 )} For the PFGE result interpretation, the Dice similarity coefficient was used, and isolates were considered identical when patterns showed ≥90% similarity.^{( 24 )} Isolates were classified as endemic or sporadic following Riley et al.^{( 25 )} Dendrograms were generated using BioNumerics v7.5 for the visualization of PFGE results.

Whole genome sequencing (WGS) was performed according to the protocol described by Migliorini et al.^{( 11 )} De novo assembly was used to determine STs through the software ‘mlst’ v2.19.0 with default settings (https://github.com/tseemann/mlst, accessed on October 20, 2021), using the K. pneumoniae scheme available via Institut Pasteur’s database (https://bigsdb.pasteur.fr/). Identification of capsule synthesis loci (K-loci or KL) and O antigen (lipopolysaccharide) serotype prediction were performed using Kleborate v.2.2.0 with default settings.^{( 26 )} Sequence assemblies were annotated using the Prokka v1.14.6 bacterial annotation pipeline with default parameters.^{( 27 )} Annotated GFF3 files were used for core genome definition using Roary v3.13.0, choosing a minimum BLASTP identity of 95% and a core gene prevalence of >99% in all isolates. The core genome alignment performed by Roary was used to infer a maximum-likelihood phylogeny using IQ-TREE v2.2.0 with 1000 ultrafast bootstrap replicates.^{( 28 )} IQ-TREE was used with the ModelFinder module, which allows the selection of the best nucleotide substitution model based on the input data. For our dataset, the best model was GTR+F+I+I+R2. The phylogenetic tree was visualized using iToL software.^{( 29 )} The clonality and relatedness of the isolates were determined using the core-genome MLST (cgMLST) analysis implemented in the chewBBACA pipeline.^{( 30 )} Briefly, the gene prediction algorithm, Prodigal, was trained using K. pneumoniae subsp. pneumoniae NTUH-K2044 (accession number: NC_012731.1). A reference cgMLST dataset for the species complex K. pneumoniae , K. variicola , and K. quasipneumoniae was downloaded from https://www.cgmlst.org/ and used for cgMLST. The resulting cgMLST data were analyzed using GrapeTree v1.5.0 to obtain a minimum spanning tree with parameters implemented in MSTree v2, ignoring missing values for the entire collection.^{( 31 )}

The resistome, virulome, and incompatibility groups were analyzed using ABRicate v.1.0.1 (ABRicate, https://github.com/tseemann/abricate). Using this tool, a blastn search of genes included in the NCBI AMRFinderPlus, VFDB, and PlasmidFinder databases was performed on the de novo whole-genome assembly.^{( 32 - 34 )} The presence of the integrative and conjugative element of K. pneumoniae (ICEKp) was identified using Kleborate v.2.2.0 with default settings.^{( 26 )}

This work was approved by the Institutional Review Board of Hospital Israelita Albert Einstein (CAAE: 39720720.7.0000.0071; # 4.404.677).

Between January 2018 and January 2021, 263 K. pneumoniae isolates were identified from the infection sites (blood, bronchoalveolar lavage, and tracheal secretions) of patients admitted to the ICU and semi-ICU, and 202 isolates of K. pneumoniae were identified in surveillance rectal swabs. The 263 isolates were distributed among the following infection sites: tracheal secretions (181 isolates, 68.8%), blood (56 isolates, 21.3%), and bronchoalveolar lavage (26 isolates, 9.9%). Of these isolates, 51.3% were susceptible to carbapenems, referred to here as non-KpCR, and the remaining isolates were carbapenem-resistant, referred to in this study as KpCR. Figure 1 summarizes the study population and filters used for the selection of isolates for molecular analyses.

Figure 1 PFGE: pulsed-field gel electrophoresis. WGS: whole-genome sequencing. KPC: Klebsiella pneumoniae carbapenemase.

For the four classes of antimicrobials tested against the non-KpCR isolates, we observed an increase in the percentage of resistant isolates from 2020 to 2021 compared to that in the 2018–2019 period ( Figure 2A and 2B). A similar trend was observed for each antimicrobial, except for amikacin, which showed the highest resistance rates in 2018 and 2021 (11.6% and 6.4%, respectively) and the lowest rates in 2019 and 2020 (3.8% and 0%, respectively). Notably, in January 2021, the last period evaluated in this study, 10 isolates were tested for amikacin, and none were resistant ( Figure 2C ). Thus, for the non-KpCR isolates, the most effective antimicrobial in vitro was amikacin, with an average susceptibility of 94.5% over the entire study period ( Figure 2B ). Meanwhile, 48.7% (128/263) of the isolates were resistant to carbapenems (imipenem and/or meropenem), referred to as KpCR. The highest number of reported KpCR isolates was 13 in January 2021 ( Figure 2D ). These isolates also exhibited high resistance to the other classes of antimicrobials evaluated in this study (Figures 2E and 2F). The analysis of isolates from surveillance rectal swabs revealed that 99.5% (201/202) were resistant to carbapenems (imipenem and/or meropenem).

Between January 2018 and January 2021, the average hospital-acquired infection (HI) rate associated with KpCR was 10.80 per 10,000 patient days. During this period, three instances exceeded the established control limit (April 2018, July 2020, and January 2021), indicating an epidemic period, with hospital infection rates of 35.8, 29.5, and 28.2, respectively ( Figure 2G ). During the pandemic, there were two instances (July 2020 and January 2021) in which the HI rate for KpCR exceeded the established control limit (29.5 and 28.2, respectively); however, these rates were lower than those in April 2018 before the pandemic ( Figure 2G ).

A total of 54 KpKPC isolates were analyzed using PFGE following the selection criteria outlined in the Methods section. These isolates were distributed across 13 points on the endemic level graph ( Figure 2G , highlighted in blue) and classified as either infection isolates (14/54; 25.9%) or colonization isolates (40/54; 74%).

Pulsed-field gel electrophoresis analysis of these 54 KpKPC isolates identified 34 distinct pulsotypes, forming 14 clusters (defined as two or more isolates with ≥90% similarity), grouping 64.8% of KpKPC isolates ( Figure 3A ). Clusters 8, 9, 11, 16, 18, and 33 included endemic KpKPC isolates, accounting for 22.2% (12/54), whereas the remaining 77.8% (42/54) were classified as sporadic. Among the infection isolates (14/54; 25.9%), 21.4% (3/14) were endemic, and 66.7% (2/3) were identified after the first COVID-19 case in Brazil. Similarly, 22.5% (9/40) of the carriage isolates obtained from surveillance rectal swabs were endemic, and 66.7% (6/9) of these were identified after the first COVID-19 case. These findings indicate that the proportion of endemic isolates remained consistent before and during the COVID-19 pandemic.

Figure 2

Specific KpKPC isolates were selected for PFGE typing during the periods when the established control limit for KpCR was exceeded following the first confirmed COVID-19 case in Brazil (July 2020 and January 2021) ( Figure 2G ). In the first period (July 2020), seven isolates were analyzed, of which five had different pulsotypes and were classified as sporadic clones. In the second period (January 2021), three isolates were analyzed, all with different pulsotypes and classified as sporadic clones. These results suggest an increase in sporadic KpKPC isolates during epidemic periods, when the control limit for KpCR was exceeded.

Between May 2018, when the earliest KpKPC isolate was analyzed by PFGE, and January 2020, the last month before COVID-19 reached Brazil, 13 isolates were typed, with 38.5% (5/13) classified as endemic and 61.5% (8/13) as sporadic. From February 25, 2020, when the first COVID-19 case was reported in Brazil, until January 2021, the last month of analysis, 41 KpKPC isolates were analyzed, of which 17.1% (7/41) were classified as endemic and 82.9% (34/41) as sporadic. These findings highlight an increase in sporadic isolates following the first COVID-19 case in Brazil, with sporadic isolates comprising 82.9% of the total isolates during this period, compared to 61.5% in the earlier period.

The 40 KpKPC isolates subjected to WGS were as follows: 26 isolates (65%) from rectal surveillance swabs, 11 from tracheal secretions (27.5%), two from blood cultures (5%), and one from bronchoalveolar lavage (2.5%) ( Table 1 ). The distribution of the 40 isolates at the endemic level is shown in Figure 2G (highlighted in red).

Table 1

Isolate	Date of isolation, month/year	Origin of specimen	MLST type (ST)	Clonal group	Plasmid type	Yersiniabactin/clb	K-Locus	O-Locus
KP728	08/2020	Surveillance rectal swabs	ST437	258	pKPC_FCF_3SP	ICEKp10/clb3	KL36	O4
KP737	07/2020	Surveillance rectal swabs	ST11	258	pKpQIL	ICEKp10/clb3	KL64	O2
KP738	07/2020	Tracheal secretion	ST437	258	pKPC_FCF_3SP	ICEKp10/clb3	KL36	O4
KP801	08/2020	Surveillance rectal swabs	ST11	258	Not detected	ICEKp3	KL15	O4
KP862	08/2020	Surveillance rectal swabs	ST16	Not applicable	pKPC_FCF_3SP	Not detected	KL51	O3b
KP863	08/2020	Surveillance rectal swabs	ST11	258	Not detected	ICEKp4	KL105	O2
KP864	08/2020	Surveillance rectal swabs	ST443	Not applicable	pKpQIL	Unknown	KL146	Unknown
KP867	09/2020	Surveillance rectal swabs	ST307	Not applicable	Not detected	Not detected	KL102	O2
KP868	09/2020	Surveillance rectal swabs	ST13	Not applicable	pKPC_FCF_3SP	ICEKp4	KL3	O1
KP869	09/2020	Surveillance rectal swabs	ST437	258	pKPC_FCF_3SP	Not detected	KL36	O4
KP871	10/2020	Surveillance rectal swabs	ST11	258	Not detected	ICEKp3	KL15	O4
KP872	12/2020	Tracheal secretion	ST11	258	Not detected	ICEKp10/clb3	KL64	O2
KP874	10/2020	Surveillance rectal swabs	ST11	258	Not detected	ICEKp3	KL15	O4
KP875	10/2020	Surveillance rectal swabs	ST11	258	Not detected	ICEKp3	KL15	O4
KP876	11/2020	Surveillance rectal swabs	ST437	258	pKPC_FCF_3SP	Not detected	KL36	O4
KP877	11/2020	Surveillance rectal swabs	ST11	258	pKPC_FCF_3SP	ICEKp3	KL15	O4
KP879	11/2020	Surveillance rectal swabs	ST11	258	Not detected	ICEKp4	KL105	Unknown
KP881	11/2020	Surveillance rectal swabs	ST11	258	pKPC_FCF_3SP	ICEKp10/clb3	KL64	O2
KP882	11/2020	Surveillance rectal swabs	ST11	258	Not detected	unknown	KL64	O4
KP884	11/2020	Surveillance rectal swabs	ST11	258	Not detected	ICEKp10/clb3	KL64	O2
KP885	01/2021	Tracheal secretion	ST16	Not applicable	pKpQIL	ICEKp10/clb3	KL51	O3b
KP886	07/2020	Tracheal secretion	ST512	258	pKpQIL	ICEKp4	KL107	O2
KP887	07/2020	Tracheal secretion	ST437	258	pKPC_FCF_3SP	ICEKp10/clb3	KL36	O4
KP888	08/2020	Tracheal secretion	ST11	258	pKPC_FCF_3SP	ICEKp3	KL15	Unknown
KP889	10/2020	Blood	ST11	258	Not detected	ICEKp3	KL15	Unknown
KP890	11/2020	Tracheal secretion	ST11	258	Not detected	ICEKp3	KL15	O4
KP891	11/2020	Tracheal secretion	ST101	101	Not detected	ICEKp3	KL17	O1
KP896	12/2020	Surveillance rectal swabs	ST11	258	Not detected	ICEKp3	KL15	O4
KP897	12/2020	Surveillance rectal swabs	ST323	Not applicable	Not detected	Not detected	KL21	Unknown
KP898	01/2021	Surveillance rectal swabs	ST307	Not applicable	Not detected	Not detected	KL102	O2
KP899	01/2021	Surveillance rectal swabs	ST11	258	pKPC_FCF_3SP	ICEKp10/clb3	KL64	O2
KP901	01/2020	Surveillance rectal swabs	ST16	Not applicable	pKpQIL	ICEKp10/clb3	KL38	O2
KP903	07/2020	Surveillance rectal swabs	ST11	258	Not detected	ICEKp10/clb3	KL15	Unknown
KP904	11/2018	Surveillance rectal swabs	ST11	258	Not detected	ICEKp3	KL15	O1
KP905	11/2019	Surveillance rectal swabs	ST101	101	Not detected	ICEKp3	KL36	O4
KP906	05/2018	Tracheal secretion	ST437	258	pKPC_FCF_3SP	ICEKp10/clb3	KL36	O4
KP908	05/2018	Bronchoalveolar lavage	ST437	258	pKPC_FCF_3SP	ICEKp10/clb3	KL36	O4
KP910	12/2018	Blood	ST11	258	pKPC_FCF_3SP	ICEKp3	KL15	O4
KP912	11/2019	Tracheal secretion	ST11	258	pKpQIL	ICEKp3	KL15	O4
KP914	10/2019	Tracheal secretion	ST443	Not applicable	pKpQIL	unknown	KL146	O1

KpKPC: Klebsiella pneumoniae carrying bla _KPC gene. MLST: multilocus sequence typing. ST: sequence type.

Multilocus sequence typing identified nine STs among the 40 sequenced KpKPC isolates, with ST11 being the most prevalent (21/40; 52.5%) ( Figure 3B ). During the periods exceeding the upper control limit (July 2020 and January 2021), four distinct STs were observed in July 2020 (ST11, ST437, ST443, and ST512), whereas only three STs (ST11, ST16, and ST307) were detected in January 2021. Before the first reported case of COVID-19 in Brazil (February 25, 2020), the detected STs included ST11, ST16, ST101, ST437, and ST443. After the first reported COVID-19 case, a greater diversity of STs was identified, including ST11, ST13, ST16, ST101, ST307, ST323, ST437, ST443, and ST512 ( Figure 3B and 3C). All STs present before the first COVID-19 case were also identified afterward, but ST13, ST307, ST323, and ST512 were exclusively detected post-COVID-19. Throughout both periods (pre- and post-COVID-19 in Brazil), ST11 predominated, accounting for 37.5% (3/8) of the isolates before the first case and 56.3% (18/32) of the isolates afterward ( Figure 3C ).

The genetic relatedness of the 40 isolates, inferred from cgMLST ( Figure 3D ), revealed high variability in the allelic profiles of cgMLST targets. The observed allelic differences ranged from 17 to 1,852 alleles. These findings were consistent with the PFGE typing results, indicating that the isolates reported during the COVID-19 period at the evaluated hospital could be classified as sporadic. No outbreak was detected during the study period, based on the threshold of ≤10 allelic differences as suggested in the literature.^{( 35 )}

Core genome analysis of the 40 KpKPC isolates identified 3,936 genes as part of the core genome (genes shared by 99–100% of the isolates), while the pan-genome comprised 10,575 genes. A phylogenetic tree was generated from the core genome alignment using Roary software ( Figure 4A ). This tree shows that the isolates were grouped by ST, with high similarity observed among isolates within each ST. The ST11 KpKPC isolates formed three main clusters, which were further subdivided based on the type of ICEKp identified in each isolate ( Figure 4A ).

Figure 3

Virulence factor typing revealed that nearly all isolates (34/40 KpKPC, 85%) carried ICEKp, and a combination of ICEKp and colibactin was present in 13 KpKPC isolates (32.5%) ( Table 1 and Figure 4A ). ICEKp4 was identified only in isolates that appeared after the first case of COVID-19, and a higher proportion of isolates carrying ICEKp3 was also observed during this period ( Figure 4B ). Both K-typing and O-typing have significant clinical and epidemiological relevance. Twelve K-types were identified, with KL15 and KL36 being the most prevalent, occurring in 32.5% and 20% of the sequenced KpKPC populations, respectively. However, only three KL types were found in both periods (before and during the COVID-19 pandemic): eight KL types were exclusively identified during the pandemic, and KL38 was found exclusively before the pandemic ( Figure 4C ). Four known O-types were identified, with O4 being the most prevalent and observed in 18 isolates (45%) ( Table 1 and Figure 4D ). Among the four identified O-types, only O3b was detected exclusively during the COVID-19 period ( Figure 4D ).

Regarding plasmid replicon typing, 23 plasmid replicons were identified among the 40 KpKPC isolates. IncFIB (K) was the most prevalent replicon, detected in 80% (32/40) of KpKPC isolates. More than 50% (12/23) of the replicons were found in at least one KpKPC isolate, both before and after the first reported COVID-19 case, whereas approximately 47% (11/23) were identified exclusively in isolates collected during the pandemic (Figure 1S, Supplementary Material ). We also evaluated whether the plasmids were similar to those reported previously at the same hospital.^{( 9 )} Fourteen KpKPC isolates (35%) harbored a plasmid highly similar to the previously reported pKPC_FCF3SP plasmid (IncN) (accession number: CP004367.2), and seven isolates (17.5%) carried a plasmid highly similar to the previously reported pKpQIL plasmid (IncFII) (accession number: GU595196.1) ( Figure 4E and 4F ). A higher proportion of isolates carrying the pKpQIL plasmid was identified before the first reported case of COVID-19 in Brazil (37.5% versus 12.9%) (Figure 2S, Supplementary Material ).

Antimicrobial resistance genes associated with 12 classes of antimicrobials were identified (Table 1S, Supplementary Material ). Except for the bla _KPC gene, the presence of which was confirmed by real-time PCR and which was part of the selection criteria for this study, no resistance genes were detected in all isolates. The most frequently reported genes before and during the COVID-19 pandemic were fosA6 (fosfomycin resistance glutathione transferase fosA6) (6/8; 75% versus 29/32; 90.6%), oqxA1 (multidrug efflux RND transporter periplasmic adaptor subunit oqxA) (5/8; 62.5% versus 24/32; 75%), bla _SHV-158(class A beta-lactamase SHV-158) (5/8; 62.5% versus 24/32; 75%), oqxB (multidrug efflux RND transporter permease subunit oqxB) (5/8; 62.5% versus 22/32; 68.8%), and sul1 (sulfonamide-resistant dihydropteroate synthase sul1) (4/8; 50% versus 21/32; 65.6%) (Table 1S, Supplementary Material ). The bla _OXA-β-lactamase gene (class D) was detected in 55% (22/40) of KpKPC isolates and was more prevalent before COVID-19 (6/8; 75%) than during the pandemic (16/32; 50%). CTX-M enzyme-related genes were also analyzed, and the bla _CTX-M-15 (extended-spectrum class A beta-lactamase CTX-M-15) gene was detected in 50% (20/40) of isolates, with a higher proportion before COVID-19 (6/8; 75%) than during the pandemic (14/32; 43.8%). The TEM-1 β-lactamase (broad-spectrum class A beta-lactamase TEM-1) gene was identified in 45% (18/40) of KpKPC isolates, with similar proportions between the two periods (50% and 43% before and during the COVID-19 pandemic, respectively). Approximately 43% (26/60) of the AMR genes were identified exclusively in isolates reported after the first case of COVID-19 (Table 1S, Supplementary Material ).