Klebsiella spp. cause severe and fatal disease in Mozambican children: antimicrobial resistance profile and molecular characterization

Background Klebsiella spp. are important pathogens associated with bacteremia among admitted children and is among the leading cause of death in children < 5 years in postmortem studies, supporting a larger role than previously considered in childhood mortality. Herein, we compared the antimicrobial susceptibility, mechanisms of resistance, and the virulence profile of Klebsiella spp. from admitted and postmortem children. Methods Antimicrobial susceptibility and virulence factors of Klebsiella spp. recovered from blood samples collected upon admission to the hospital (n = 88) and postmortem blood (n = 23) from children < 5 years were assessed by disk diffusion and multiplex PCR. Results Klebsiella isolates from postmortem blood were likely to be ceftriaxone resistant (69.6%, 16/23 vs. 48.9%, 43/88, p = 0.045) or extended-spectrum β-lactamase (ESBL) producers (60.9%, 14/23 vs. 25%, 22/88, p = 0.001) compared to those from admitted children. blaCTX-M-15 was the most frequent ESBL gene: 65.3%, 9/14 in postmortem isolates and 22.7% (5/22) from admitted children. We found higher frequency of genes associated with hypermucoviscosity phenotype and invasin in postmortem isolates than those from admitted children: rmpA (30.4%; 7/23 vs. 9.1%, 8/88, p = 0.011), wzi-K1 (34.7%; 8/23 vs. 8%; 7/88, p = 0.002) and traT (60.8%; 14/23 vs. 10.2%; 9/88, p < 0.0001), respectively. Additionally, serine protease auto-transporters of Enterobacteriaceae were detected from 1.8% (pic) to 12.6% (pet) among all isolates. Klebsiella case fatality rate was 30.7% (23/75). Conclusion Multidrug resistant Klebsiella spp. harboring genes associated with hypermucoviscosity phenotype has emerged in Mozambique causing invasive fatal disease in children; highlighting the urgent need for prompt diagnosis, appropriate treatment and effective preventive measures for infection control. Supplementary Information The online version contains supplementary material available at 10.1186/s12879-021-06245-x.


Background
Klebsiella spp. are among the most frequently reported pathogens associated with opportunistic infections in immunocompromised individuals and patients hospitalized for prolonged periods [1], and commonly implicated in nosocomial urinary tract infections and sepsis in infants from sub-Saharan Africa [2,3]. Given the importance of Klebsiella spp. in nosocomial infection and the continuous increase and spread of communityacquired multi-drug resistant (MDR) Klebsiella pneumoniae, including Extended-Spectrum β-Lactamase (ESBL) producers, the associated high mortality is causing a grave concern [2,4].
Recently, postmortem studies conducted in urban (Maputo city) and rural (Manhiça District) Mozambique to ascertain causes of death in stillbirths and young children using minimally invasive tissue sampling (MITS) approaches found Klebsiella spp. to be among the top pathogens assigned in the chain of events leading to death [14][15][16].
Data on the molecular epidemiology and characterization of Klebsiella spp. are scarce in Africa including Mozambique [17][18][19]. However, data from an urban hospital in the city of Beira, central Mozambique showed high frequency (100% of isolates) of ESBL-producing K. pneumoniae among hospitalized children with urinary tract infections [20]. Additionally, in our previous work we documented the occurrence of community-acquired Klebsiella spp. strains harboring bla CTX-M-15 among children living in the Manhiça District, southern Mozambique [21,22]. Herein, we described and compared the molecular characterization, antimicrobial susceptibility of Klebsiella spp. isolates recovered from bloodstream infections and postmortem blood specimens collected in children under 5 years of age and we characterize the MDR associated Klebsiella spp. from admitted children with bloodstream infections who have died. In this prospective study, we documented the occurrence of hypermucoviscous, multidrug-resistant Klebsiella spp. with an ESBL-producing phenotype, harboring bla CTX-M- 15 , and causing invasive disease associated with poor outcomes in infants and young children in a rural hospital in Southern Mozambique.

Clinical isolates and study area
We analyzed 111 Klebsiella spp. isolates recovered from bloodstream infection (n = 88) of children under 5 years old admitted at the Manhiça District Hospital (January 2001 and December 2019) and from postmortem blood specimens (n = 23) of deceased children under 5 years old enrolled in the ongoing Child Health and Mortality Prevention Surveillance (CHAMPS) project (December 2016 and December 2019) [23]. The CHAMPS Network was established in seven countries in south Asia and sub-Saharan Africa including Mozambique (Manhiça site) to collect standardized, population-based, longitudinal data on mortality among children under 5 years and stillbirths, to improve the accuracy of causes of death determination using a novel approach of minimally invasive tissue sampling (MITS) [24].
In Mozambique, CHAMPS is being conducted by the Centro de Investigação em Saúde de Manhiça (CISM) in the district of Manhiça, a rural area of Maputo Province in southern Mozambique, with an estimated population of approximately 201,845 inhabitants. In this site, a continuous demographic surveillance system (DSS) has been running since 1996 with regular update of vital events; and currently covering the entire district of Manhiça [25].
In addition to the DSS, CISM has been conducting morbidity surveillance which includes invasive bacterial disease surveillance in the pediatric population (children < 15 years of age) at the Manhiça District Hospital (MDH) (a 150 beds referral district hospital of the Manhiça district) and other 5 peripheral health facilities since January 1997, documenting all outpatients and inpatients visits using standard questionnaires. Blood cultures were routinely performed as part of the clinical practice upon hospital admission for all children under 2 years old and for older children under 15 years with axillary temperature ≥ 39°C or who meet, according to the admitting clinician, specific severity criteria [4,26]. Bacterial cultures were performed at the CISM Microbiology laboratory for Klebsiella spp. and other bacterial isolation and identification using conventional microbiology and biochemical methods [4]. Clinically relevant isolates were stored at − 70°C in a biobank (in a STGG medium -2% of Skim milk + 3% tryptic soy broth + 0.5% glucose + 10% glycerol) for further characterization [27]. Klebsiella bacteremia was defined as Klebsiella positive blood culture collected from children upon admission at the MDH.
Under the health demographic surveillance system (HDSS) platform, CISM has been conducting a mortality surveillance in the context of the CHAMPS network, since 2016. CHAMPS uses the MITS approach on deaths reported from the community or health facility, to track the definitive/immediate causes of deaths in stillbirths and children under 5 years old, as described elsewhere [14,23]. Tissue samples, including blood and cerebrospinal fluid (CSF) are investigated using conventional microbiology and molecular techniques for pathogens detection, including Klebsiella [23,28,29]. The causes of death were determined by a panel of experts that comprises at least 1 of each of the following: clinician (eg, pediatricians, neonatologists, and obstetricians), pathologist, epidemiologist, and microbiologist. The Panelists receive all available information from linked maternal data, child clinical data, individual demographic data, verbal autopsy, microbiology, molecular testing, clinical diagnostics (human immunodeficiency virus [HIV], tuberculosis, malaria), photographs of the deceased from the MITS procedure, and histopathology findings in the form of a determination of cause of death package and assign the underlying, antemortem, and immediate (and other contributing) causes of death [14]. Postmortem Klebsiella is defined as Klebsiella isolates recovered from blood samples collected through MITS in deceased children enrolled in the CHAMPS.
The selection of the antibiotics tested was done according to their use for local empirical treatment and those recommended by the Clinical Laboratory Standard Institute (CLSI) guidelines [30]. The interpretative category of resistance was determined according to the CLSI guidelines (29th Edition) [30]. All isolates resistant to three or more unrelated families of antibiotics were defined as MDR [31]. In the MDH, children are treated empirically with ampicillin or ampicillin + gentamicin and amikacin or ceftriaxone.

Detection of ESBL phenotype and genotype
All ceftriaxone, aztreonam or ceftazidime resistant isolates were then screened for ESBL production by disk synergy assay with the addition of a beta-lactamase inhibitor disc, as previously described [32]. Briefly, AMP, CAZ, and CRO disk were placed 2 cm distant and AUG disk was placed on the center closer to the other disks on Muller-Hinton agar plate inoculated with 0.5McF of Klebsiella isolates. The tests were considered positive if lightly growth-inhibitory zone (phantom zone) was observed between the disks. Isolates with ESBL phenotype were assessed by a multiplex polymerase chain reaction (PCR) for bla CTX-M , bla TEM , bla SHV genes, using universal primers ( Table 1). The multiplex PCR was performed in 25 μL reaction volumes, including 0.5 μM of each primer, 12,5 μL of 2X master mix (Fermentas, Thermo Scientific, MA, USA), 3 μL of genomic DNA and nuclease free water. Cycling conditions included an initial denaturation step of 5 min at 95°C, followed by 30 amplification cycles (denaturation at 94°C for 1 min, annealing at 57°C for 1 min and elongation at 72°C for 2 min) and a final extension step at 72°C for 10 min.
Amplicons of the expected size, for bla CTX-M , bla TEM , bla SHV genes were purified using QIAquick PCR purification kit (QIAGEN, Hilden, Germany), and shipped to a commercial sequencing services company (Macrogen Europe BV, Amsterdam, the Netherlands) for Sanger sequencing in both directions using the same sets of primers for each gene (Table 1). Sequence analysis and alignments were performed using the BioEdit software (v7.2.5) and the BLAST Program (http://blast.ncbi.nlm. nih.gov/Blast.cgi), which was also used to determine the allele correspondent for each gene (bla CTX-M , bla TEM , bla SHV ).

Detection of virulence genes
The virulence genes associated with protectins and invasins (rmpA, wzi-K1, and traT), adhesins (fimH, mrkD, and ycfM), toxins (hylA and hylC) and siderophores (iutA, fyuA, and entB) were assessed by 4 multiplexed PCRs for each group of genes, optimized in this study and using primers previously described (Table 1). All genes, except those encoding Serine Protease Autotransporters of Enterobacteriaceae (SPATEs), were amplified in 25 μL reaction volumes, including 0.5 μM of each primer, 12,5 μL of 2X master mix (PCR Master Mix 2x, Thermo Scientific, MA, USA), 3 μL of genomic DNA and nuclease free water. The cycling conditions included an initial denaturation step at 95°C for 2 min, followed by 35 amplification cycles with denaturation at 94°C for 30 s, annealing for 1.5 min (temperatures are described in Table 1) an elongation at 72°C for 1.5 min and a final extension step at 72°C for 10 min.
One PCR product of the expected size per SPATE gene (sat, sepA, sigA, piC, and pet) was also shipped for Sanger sequencing with the corresponding sets of primers (described in Table 1) for confirmation.

Hypermucoviscous phenotype
To verify the concordance with the genotype and the expression of the hypermucoviscous phenotype, all rmpA and wzi-K1 positive isolates were tested by string tests as previously described elsewhere [8].

Statistical analysis
PCR and antibiotic susceptibility data were entered in an Excel spreadsheet and checked for consistency by an independent laboratory technician. Data were analyzed using Stata software version 14.0 (STATA Corporation, College Station, Tx, USA). Frequencies and proportions were used to describe categorical variables, and differences of proportions were compared using the chisquare or Fisher's exact test where appropriate, with a pvalue of < 0.05 being considered significant. Surviving children or Srv. Bacteremia were defined as children hospitalized at the MDH who tested positive for Klebsiella spp. in blood culture upon admission and that were alive at discharge (n = 41) and dead children or Dth. Bacteremia were defined as children hospitalized at the MDH who tested positive for Klebsiella spp. in blood culture upon admission who died and had clinical information available n = 23).

Outcome of Klebsiella infections in bacteremic patients
Of the admitted children, outcome data was available for 75 out of 88 children (85.2%) with Klebsiella bacteremia, yielding an in-hospital case fatality rate (CFR) of 30.7% (23/75), which was particularly high (43.5%; 10/23) among younger than 2 years (between 12 and 23 months), and 11 children were transferred or left the hospital.

Discussion
This is the first report comparing the antimicrobial susceptibility profile, its associated mechanisms of resistance, and the virulence repertoire between Klebsiella spp. isolates from postmortem studies and isolates recovered from children < 5 years of age with severe disease. The finding of a higher prevalence of MDR strains harboring both ESBL and genes associated with the hypermucoviscous phenotype in postmortem isolates may suggest the potential virulence and contribution of specific strains of Klebsiella to life-threatening infections, and thus, its potential role in childhood mortality in Mozambique.
The high CFR observed among children admitted with bacteremia may be explained by the high proportion (66%) of MDR strains. Most MDR isolates were also resistant to ceftriaxone due to the presence of bla CTX-M-15 in more than 25% of isolates limiting the therapeutic options, which is consistent with our previous observations of the poor outcome of Klebsiella associated bacteremia in children younger than 14 years of age [4,22,40]. The  low resistance to chloramphenicol among postmortem isolates may be explained by the fact that this antibiotic was used as a drug of choice for empirical therapy [4] and is no longer a choice for routine clinical management. On the other hand, the presence of hylC and sat among the fatal strains from bacteremic children may also help to explain the Klebsiella associated mortality, as they have been described in E.coli showing cytotoxic effect on erythrocytes and epithelial cells, respectively, by disrupting the membrane or autophagy and cell detachment [41][42][43]. The excess of Klebsiella associated mortality is similar to previous studies showing mortality rates up to 38.5% in other countries [44,45], including in the neighboring country, South Africa (26-35%) [46]. As previously reported in this community and elsewhere [10,17,22,47,48], an important public health concern is related to the emergence and widespread dissemination of ESBL-producing Klebsiella spp., challenging the appropriate management of MDR Klebsiella infections, that were mostly treated with ceftriaxone. The major dilemma is related to the finding of strains with reduced susceptibility to carbapenems (meropenem and ertapenem), often used as a second-line treatment options, as per international guidelines, but not foreseen in our setting [49]. This is alarming as these antibiotics are not routinely used in our study community and are rare in Mozambique in general, even though, the isolates remained susceptible to imipenem. Our findings provide a clear indication that an appropriate action plan is required to contain the spread of antibiotic resistance, mainly because the isolates remained susceptible to antibiotics that are expensive and unavailable in most of the rural communities, constraining their use even when the available antibiotherapy is ineffective [50].
The high frequency of genes involved in the synthesis of capsule (rmpA and wzi-K1), known to inhibit and evade phagocytosis by host cells and to neutralize the antibacterial activity of host defenses [13], in postmortem isolates may suggest potential virulence and fatality of the strains recovered in deceased children. The presence of these genes, mainly, rmpA give an indication that hipervirulent Klebsiella may be circulating in our setting, as this gene is one of the markers for hvKp [8,9,51]. However, this should be interpreted with caution, as The high prevalence of siderophores was expected among pathogens causing invasive disease as these are essential for iron acquisition and survival of Klebsiella when interacting with host cells [12]; in addition to the high prevalence of adhesins that are critical for the first step of bacterial colonization /adhesion, with several variants that have been described [10,11].
As of this writing, to our best knowledge this is the first report on the occurrence of the studied SPATEs in Klebsiella spp., suggesting that they potentially play an important role in the pathogenicity of Klebsiella, as previously described [52][53][54]. In addition, the sequences of Table 4 Resistance and virulence profile of isolates from bacteremic children who died after admission SPATEs reported here are similar to those reported in other Enterobacteriaceae species as previously reported and may play the same role [52,54]. There are some limitations in this study, one of these is that our study compared Klebsiella spp. in admitted patients and postmortem samples from different study periods; none of the children with bacteremic Klebsiella isolates was included in the postmortem study, even in the period when the IBD surveillance and CHAMPS overlapped. In deep molecular characterization including phylogenetic analysis, serotyping and multi-locus sequence typing are underway to better understand the molecular epidemiology of isolates from bacteremia and postmortem samples. This data cannot be generalized for the Mozambican or African population, as the samples were from only one geographic setting.

Conclusion
Our data calls for the urgent need for more effective alternative antibiotics for management of fatal virulent MDR bla CTX-M-15 and bla SHV-1 Klebsiella strains and the implementation of appropriate measures to contain AMR (e.g. prudent use of antibiotics for empirical treatment, reduce the antibiotic usage in the agricultural activity, improve hospital infection control, rationalize the antibiotic use in the community). Continuous monitoring and surveillance for antimicrobial resistance (AMR) are crucial to support defining local and international guidelines for effective alternative options for anti-biotherapy or guide the design and development of new intervention tools (e.g. vaccines with broad coverage spectra), which requires the description of serotypes circulating. Authors' contributions AM contributed to the literature review, data analysis, conceptualization and writing up the manuscript. AM did also all the tables and Fig. 1. MG and AJ contributed equally, and the sequence of names were decided according to published papers. AJ did the Fig. 2A and B. NN contributed to all data acquisition. NB, AS, RV, SM, AC, JC, JO, HM, TN, RB, CW, DB, and QB reviewed critically the manuscript. IM contributed to the conceptualization, critically reviewed the manuscript and designed the study. All authors read and approved the manuscript.

Availability of data and materials
The datasets used and analysed during the current study are available from the corresponding author on reasonable request.

Declarations
Ethics approval and consent to participate The isolates included in this research were obtained from participants who were enrolled in the IBD and CHAMPS, and only if the parents, caretaker or legal guardians accepted to participate and sign an informed consent. The IBD and CHAMPS protocol and its methods were performed in accordance with the Declaration of Helsinki and Good Clinical and Laboratory Practices.

Consent for publication
Not applicable.

Competing interests
The authors declare that they have no competing interests.