High third-generation cephalosporin resistant Enterobacteriaceae prevalence rate among neonatal infections in Dakar, Senegal

Background Neonatal infection constitutes one of Senegal’s most important public health problems, with a mortality rate of 41 deaths per 1,000 live births. Methods Between January 2007 and March 2008, 242 neonates with suspected infection were recruited at three neonatal intensive care units in three major tertiary care centers in Dakar, the capital of Senegal. Neonatal infections were confirmed by positive bacterial blood or cerebrospinal fluid culture. The microbiological pattern of neonatal infections and the antibiotic susceptibility of the isolates were characterized. In addition, the genetic basis for antibiotic resistance and the genetic background of third-generation cephalosporin-resistant (3GC-R) Enterobacteriaceae were studied. Results A bacteriological infection was confirmed in 36.4 % (88/242) of neonates: 22.7 % (30/132) during the early-onset and 52.7 % (58/110) during the late-onset periods (p > 0.20). Group B streptococci accounted for 6.8 % of the 88 collected bacterial isolates, while most of them were Enterobacteriaceae (n = 69, 78.4 %). Of these, 55/69 (79.7 %) were 3GC-R. The bla CTX-M-15 allele, the bla SHV and the bla TEM were highly prevalent (63.5, 65.4 and 53.8 %, respectively), usually associated with qnr genes (65.4 %). Clonally related strains of 3GC-R Klebsiella pneumoniae and 3GC-R Enterobacter cloacae, the two most commonly recovered 3GC-R Enterobacteriaceae (48/55), were detected at the three hospitals, underlining the role of cross-transmission in their spread. The overall case fatality rate was 18.6 %. Conclusions Measures should be taken to prevent nosocomial infections and the selection of resistant bacteria.


Background
The fourth Millennium Development Goal, established after the United Nations Millennium Summit in 2000, seeks to decrease the mortality of children under five years of age by two-thirds before 2015 [1]. Sub-Saharan Africa has seen a rapid decline in its under-five mortality rate, with the annual rate of reduction doubling between 1990-2000 and 2000-2013. However, progress in reducing the neonatal mortality rate remains virtually stagnant in this part of the world. The number of neonatal deaths has in fact increased in sub-Saharan Africa, from 944,000 in 1990 to 1,029,000 in 2013 [2].
Neonatal sepsis is defined as a clinical syndrome occurring within the first 28 days of life, manifested by systemic signs of infection, and isolation of a bacterial pathogen from the bloodstream. Neonatal bloodstream infection is associated with significant morbidity, poor long-term outcomes, and a reported mortality rate of 12-20 % [3,4]. The early-onset period, defined as infection within three days of birth (except for group B streptococci (GBS), 0-6 days after birth), classically reflects vertically transmitted infections. Late-onset infections are usually attributed to bacteria acquired from the infant's surroundings (hospitals or community), Klebsiella spp. and Staphylococcus aureus being the most frequently recovered [5]. Among hospital born babies, most neonatal infections are hospitalacquired (HA) in developing countries, irrespective of the period of infection. Extended-spectrum beta-lactamase (ESBL)-producing Klebsiella pneumoniae is the most frequent HA pathogen [5], along with Staphylococcus aureus and Escherichia coli [6]. Third-generation cephalosporins (3GC) are commonly used as first-line treatment for neonatal infections in developing countries, resulting in a gradual upward trend in the number of antibiotic resistant isolates, especially among Enterobacteriaceae [5].
Senegal is a resource-limited country in equatorial Africa (ranked 163/187 according to the Human Development Index in 2013). Neonatal infection is one of Senegal's most important public health problems, with a neonatal mortality rate of 21 deaths per 1,000 live births in 2015 [7]. This is four times the rate observed in developed countries [8]. Moreover, there is little data on bacterial resistance patterns in neonatal infections in Senegal or in developing countries [6].
We characterized the bacterial pathogens associated with neonatal infection and their antibiotic susceptibility patterns, in three major tertiary care centers; we also studied the genetic basis for antibiotic resistance and the genetic background of third-generation cephalosporins resistant (3GC-R) Enterobacteriaceae.

Patients
From January 2007 to March 2008, neonates (≤28 days) admitted to the neonatal intensive care units in the Hôpital Principal de Dakar (HPD), Hôpital Aristide Le Dantec (HALD) and Hôpital des Enfants Albert Royer (HEAR), were consecutively pre-included if infection was suspected according to the presence of at least two of the following clinical signs: fever (temperature > 37.8°C) or hypothermia (temperature <35°C), respiratory signs (apnea, tachypnea (>60 breaths per minute), respiratory distress), neurological signs (hypotonia, weak reflexes, refusal to suck, perturbation of consciousness, convulsions, coma, irritability), digestive signs (vomiting, diarrhea), jaundice (early [<24 h after birth] or prolonged) and cutaneous signs (purpura). One blood culture (standard volume 1-2 ml) was performed before starting antibiotic treatment for each neonate. A lumbar puncture was prescribed except i) in severely ill infants with either cardiovascular or respiratory distress, tense or bulbing anterior fontanelle, the presence of severe thrombocytopenia, or infection around the lumbosacral region [9], ii) in case of parents or legal guardians's refusal. Neonatal infection was confirmed by a positive blood and/or cerebrospinal fluid culture. Empirical antibiotic therapy was considered inappropriate if the treatment regimen did not include at least one antibiotic that was active in vitro against the infecting microorganisms. The cause of death for patients who died during hospitalization was recorded according to the clinician's assessment. Attributable mortality was defined as neonatal death within three days after the onset of infection.
Neonates were classified into two groups depending on the age at the onset of clinical signs: early-onset period in infants between zero and three days of age (except for GBS, 0-6 days) and late-onset period in infants after three days of age (except for GBS, >6 days).

Microbiological analysis, antimicrobial susceptibility testing and detection of ESBL
Blood cultures (Hemoline performance diphasic medium, bioMerieux, Marcy l'Etoile, France) were incubated aerobically at 37°C and observed daily during seven days. In cases of visible microbial growth, blood and chocolate agar plates (Biorad laboratories, Richmond, CA) were inoculated, incubated in a 5 % CO2 atmosphere at 37°C, and examined after 24 and 48 h. Cerebrospinal fluid (CSF) were collected in sterile containers and processed within 2 h. Schaedler broth with vitamin K3, blood and chocolate agar plates (Biorad laboratories, Richmond, CA) were inoculated, incubated at 37°C in a 5 % CO2 atmosphere, and examined after 24 and 48 h. API biochemical galleries (API, bioMerieux, Marcy l'Etoile, France) were used for bacterial identification. Growth of Bacillus, Corynebacterium, and Micrococcus species as well as coagulasenegative staphylococci in the blood cultures was considered to be due to contaminants and these cultures were excluded.
Antibacterial drug susceptibility was determined using the disc diffusion method according to the guidelines of the French Society of Microbiology. Production of ESBL in 3GC-R Enterobacteriaceae was detected using the double-disc synergy test [10].
DNA extraction and detection of beta-lactamase and plasmid-mediated quinolone resistance genes Previously described PCRs were used to screen for carbapenem-hydrolyzing β-lactamase genes (bla VIM , bla IMP and bla KPC ), as well as plasmid-encoded bla CTX-M , bla AmpC , bla SHV , bla TEM beta-lactamase genes, the aac(6′)-Ib gene, and quinolone resistance qnrA/B/S and qepA genes [11][12][13]. All aac(6′)-Ib positive strains were further analyzed by digestion of the PCR product with BtsCl (New England Biolabs, Ipswich, MA) to identify the mutated aac(6')-Ib-cr gene, which lacks the BtsCl restriction site present in the wild-type gene [14]. Isolates resistant to cefoxitin were tested for the presence of six families of the plasmid-borne ampC gene (FOX, ACC, EBC, MOX, CIT, DHA) [15] and bla CMY-2 .

Pulsed Field Gel Electrophoresis of Enterobacteriaceae
Escherichia coli, Klebsiella pneumoniae and Enterobacter cloacae were analyzed by Pulsed Field Gel Electrophoresis (PFGE) as previously described [16].

Multilocus sequence typing of Group B streptococci
Group B streptococci (GBS) were studied by multilocus sequence typing (MLST) using the MLST website for assignments [17].

Data analysis
Qualitative data were compared using the chi square test and Student's t test. P values lower than 0.05 were considered to be significant. Statistical analysis was performed using Stata® software 12.0 (Stata Corporation, College Station. Tex.).

Pulsed-field Gel Electrophoresis of 3GC-R Enterobacteriaceae
The 33 K. pneumoniae isolates grouped into five main PFGE clusters (3 to 11 isolates) in Pulsed-field Gel Electrophoresis (PFGE) analysis. The 15 E. cloacae isolates yielded 10 distinct PFGE patterns, with six strains belonging to a single cluster. These six isolates were isolated in all the three hospitals. Each of the four Escherichia coli isolates displayed a distinct PFGE pattern ( Table 2).

MLST of GBS
Five of the six isolates were investigated: ST17 was isolated from two CSF samples from different hospitals, and ST23 (n = 2) and ST26 (n = 1) from three blood samples.   [18][19][20]. However, negative blood culture does not rule out a bacterial infection, as obligate anaerobes were not investigated in our study. In addition, coagulasenegative staphylococci, Bacillus spp., Corynebacterium spp. and Micrococcus spp. were excluded from this study due to the lack of two successive blood cultures, but it is crucial to recognize that they can also represent a true cause of lateonset infection. A lumbar puncture was performed only in 16 neonates with suspected neonatal infection. Disruption of African hospitals, combined with refusal by children's parents or legal guardians due probably to a cultural preexisting belief that this medical procedure will cause death [21], lead to a probable underestimate of the true prevalence of meningitis in our studied population. The overall lethality rate was 18.6 %. This value was in the range of reported values in developing countries [5], including Senegal (19.5 % in 1999) [22]. The pattern of pathogens associated with neonatal infections was quite different from that observed in developed countries were GBS and E. coli are predominant [9]. GBS and E. coli were responsible of 6.8 and 9.1 % of neonatal infection in the current study, respectively. These data are in the range of reported values in Sub-saharan Africa, 2-8 % of cases for GBS and 5-16 % for E. coli [23]. The microorganisms most frequently recovered were Enterobacteriaceae (78.4 %), for both early and late-onset infections, most of them being resistant to 3GC [5]. This indicates that hospital environmental and human floras are reservoirs of ESBL-Enterobacteriaceae and are potential sources of colonization and infection in newborns during delivery and the post-natal period. The usual conditions prevailing in hospital settings in low-income countries may favor horizontal transmission of multiresistant bacteria [24]. They probably act as a barrier for GBS and E. coli transmission, which could explain their low detection rates among infected neonates. As a result, the majority of early-onset infections may be more hospital acquired, than vertically transmitted. Of note, pregnant women may represent a substantial source of multiresistant Enterobacteriaceae for vertical transmission to neonates as highlighted by the high prevalence of ESBL-E among pregnant women from Madagascar [24]. Further studies are needed to differentiate maternally-acquired from hospital-acquired routes of transmission to implement improved preventive strategies in these settings.
Among Enterobacteriaceae, 79.8 % were 3GC-R, as previously described in developing countries [5]. Then, a significant proportion of the causal bacteria was not treatable neither by the recommended first-line regimen (ampicillin and gentamicin) [25] nor by alternative C3G treatment. Thus, carbapenems are regarded as the ultimate treatment option. However, these antimicrobials are expensive and unaffordable for the vast majority of the population in these settings, making these bacteria difficult to treat. In South-East Asia, many newborn babies in hospitals are now treated with carbapenems as firstline treatment for sepsis or suspected sepsis. However, carbapenems should not be prescribed as empiric therapy as highlighted by the emergence of resistant untreatable carbapenem-resistant Enterobacteriaceae infections associated with high mortality in intensive care units [26]. The alarming use of 3GC observed in our study, contrary to recommendations of antibiotic use [25], has contributed to selection pressure that has provided a competitive advantage for C3G resistant strains. Qualityassured basic microbiology services and routine bacterial resistance surveillance are urgently needed in developing countries in order to reduce antibiotic selection pressure [27]. In addition, all necessary measures for preventing hospital-acquired infections must be implemented. These include standard hygiene practices, especially adherence to hand hygiene policies, which are the cornerstone for preventing the transmission of multidrug-resistant bacteria [27]. Indeed, cross-transmission plays a major role, as suggested by the fact that neonates from the same department shared clonally related strains. This was true for the K. pneumoniae and E. cloacae strains, which are known to be highly transmissible in hospital settings, especially in intensive care units [26]. In addition, blood culture is time-consuming. Up to 6 days are usually required before antimicrobial susceptibility testing results of bacterial pathogens are available. This delay often causes an inappropriate empiric therapy but also a prolonged length of hospital stay, which favor the horizontal transfer of multidrug resistant strains. Neonates in different hospitals also shared clonally related strains which suggests that there was cross-transmission events between the three health care centers, probably by transfer of infected or colonized patients.
3GC-R resistance in Enterobacteriaceae is mainly mediated by ESBL genes located on mobile genetic elements such as plasmids [28]. In our study conducted in 2007-2008, bla CTX-M group 1 or bla SHV was detected in most of the strains (69.5 %), highlighting the bla TEM / bla SHV to bla CTX-M-15 shift in the 2000s , observed in Dakar [10,29] and worldwide [30]. These genes were frequently associated with qnr and aac(6')-Ib-cr genes. Frequent association of bla CTXM-15 , qnrB and aac(6')-Ib-cr genes has been described in Enterobacteriaceae isolates in North, West, Central and East Africa, consistent with the hypothesis that these resistance-determinant genes are carried on the same plasmid [10,29,31]. Such an accumulation of resistance genes on a single plasmid prone to horizontal transmission is problematic, especially in countries with inadequate health care systems. Early and appropriate antibiotic treatment is the cornerstone of neonatal infection management [32]. This is of great concern, as 65.9 % of neonates did not receive an appropriate empiric therapy in our study. The emergence of carbapenemases, which has been described in Senegal [33] but was not found in our study, could be a serious challenge for infection control and antibiotic therapy in the future.
In our study, there were only six cases of mother-tochild transmission of GBS. Two isolates, both found in infants with meningitis, were assigned to the ST17 hypervirulent lineage. This ST is known to be associated with neonatal disease and correlated with serotype III [34,35] which is predominant in meningitis or invasive infections [34]. In 2007, a study conducted in Dakar on healthy pregnant Senegalese women showed a GBS positive vaginal prevalence rate of 20 %. The three STs observed in our study were all major clones in the previous study; ST17, ST23 and ST26, representing 13.3, 14.7 and 20.0 %, respectively, of all GBS in cases of vaginal carriage [36]. Given the resources required for a screeningbased approach to guide intrapartum antibiotic prophylaxis, a risk factor-based approach with antibiotics prescribed during labor for women showing risk factors of early-onset GBS disease such as chorioamnionitis, prolonged rupture of membrane, or preterm delivery, [37] is likely to be more applicable.

Conclusions
These data are of particular importance because of the difficulty to carry out such studies in countries with inadequate health care systems, and the lack of data on pathogens associated with neonatal infections and their antibiotic susceptibilities. 3GC-R Enterobacteriaceae were highly prevalent and GBS were detected. All necessary measures should be taken in African hospitals to prevent nosocomial infections and the selection of resistant bacteria by implementing efficient nosocomial infection surveillance programs. Standard hygiene and especially hand hygiene policies need to be reinforced to prevent transmission of multidrug-resistant bacteria.