High rate of antibiotic resistance among pneumococci carried by healthy children in the eastern part of the Democratic Republic of the Congo

From January 2014 to June 2015, 794 healthy children aged one to 60 months attending one of seven health centres in the South-Kivu province in the eastern part of DR Congo for immunisation or growth monitoring were included in the study. The health centres were located in the city of Bukavu (n = 3), in the suburban area (n = 1), or in the surrounding rural area (n = 3) (Additional file 1).

Written questionnaires about immunisation status and demographic factors were completed by trained final-year medical students or nurses in the presence of a paediatrician and a basic physical examination of the children was performed to monitor current signs of a respiratory tract infection. When available, the immunisation card was checked to confirm the vaccination status of the child. For the 284 healthy children enrolled in 2015, another questionnaire containing questions about socio-economic conditions and previous illness was added. The weight and height were measured and standardised for age using the Emergency Nutrition Assessment (ENA) software 2011 [27].

Signed informed consent was obtained from the parent or guardian of each included child. The study was approved by the Ethics Committees at the Université Catholique de Bukavu, DR Congo, and at the University of Gothenburg, Sweden.

A nasopharyngeal specimen was obtained from all participating children using an Eswab (Copan Diagnostics Inc., Murrieta, CA). A single trained investigator at each centre obtained the sample following a standardised procedure. The head of the child was tipped backwards and gently immobilised. The bent swab was inserted into the nostril and then passed into the nasopharynx to a distance equal to that from the nose to the tip of the ear and kept in that position for 5 s. The samples were shipped to the Clinical Laboratory at Panzi Hospital within two to 6 h for subsequent pneumococcal culture.

The samples were cultured for Streptococcus pneumoniae on 5% human blood agar plates (Oxoid Columbia Blood Agar Base – Thermo Fisher Scientific, Waltham, MA), incubated overnight at 34–36 °C in closed jars (Oxoid Limited, Thermo Fisher Scientific, Hampshire, UK) supplied with CO₂ paper sachets (BD GasPak™ EZ CO₂ Container System, Becton, Dickinson and Company, Franklin Lakes, New Jersey) and CO₂ indicators (BD CO₂ Indicator 0.5 mL, Becton, Dickinson and Company).

Suspected pneumococci were identified by a positive optochin test (diameter ≥ 14 mm) and were further tested for antibiotic susceptibility using a disc diffusion test against oxacillin (1 μg) (screening for beta-lactam resistance), trimethoprim-sulphamethoxazole (TMP-SMX) (1.25/23.75 μg), norfloxacin (10 μg) (screening for fluoroquinolone resistance, i.e. levofloxacin and moxifloxacin), tetracycline (30 μg), erythromycin (15 μg) and clindamycin (2 μg) (all from Oxoid Limited), using breakpoints according to the European Committee on Antimicrobial Susceptibility Testing, 2017 [28]. The bacteria and antibiotic discs were applied to Muller Hinton agar plates (Oxoid Limited) supplied with 5% sheep blood (Thermo Fisher Scientific) and 20 mg/L β-Nicotinamide adenine dinucleotide (NAD) (AppliChem GmbH, Darmstadt, Germany) that were incubated over night at 34–36 °C in a CO₂ environment as described above. Pneumococcal isolates with reduced sensitivity to oxacillin (diameter < 20 mm) were further tested using minimal inhibitory concentration (MIC) determination against penicillin G, ampicillin and ceftriaxone (all 0.016–256 μg/mL, bioMérieux, Marcy l’Etoile, France). Pneumococci with an MIC of > 0.06 mg/L were defined as having reduced susceptibility to benzylpenicillin. Multi-drug resistant (MDR) isolates were defined as those that were non-susceptible (intermediate or resistant) to at least one drug from three or more different classes of antimicrobial agents, including the beta-lactams (the penicillins benzylpenicillin and ampicillin and the cephalosporin ceftriaxone) [29]. Apart from the beta-lactams, all drugs tested belong to different classes, namely fluoroquinolones (norfloxacin), lincosamides (clindamycin), macrolides (erythromycin), folate pathway inhibitors (trimethoprim-sulphamethoxazole) and tetracyclines (tetracycline) [29]. The isolates were frozen (− 20 °C) in STGG storage medium [30] before being transported to Gothenburg, Sweden, for further analyses.

Out of the 163 pneumococcal strains isolated at the Clinical Laboratory at Panzi Hospital in Bukavu, 151 isolates were transported frozen in STGG medium to Gothenburg. Of these, 32 isolates could be re-cultured after storage and transport, and were tested for antibiotic susceptibility at the Department of Infectious Diseases, University of Gothenburg, Sweden, as well (Fig. 1). When the results were compared with those obtained at the laboratory in Bukavu for the same isolates, the diameter zones for all the tested antibiotic discs varied by 6 mm or less in at least 75% of the cases (Table 1). The resulting interpretation into Sensitive (S), Intermediate (I) or Resistant (R) was similar in both groups (Table 1).

The distributions of MIC values were also compared between the analyses performed in Bukavu and Gothenburg, respectively (Additional file 2). There was an even distribution of MIC values between the two sites for all of penicillin G, ampicillin and ceftriaxone. When interpreting the MIC values for penicillin G, all the isolates were categorised in the same SIR category. For ampicillin, one isolate was differently categorised into sensitive and intermediate, respectively, and the same was true in two cases for ceftriaxone. We also compared the MIC distributions between 2014 and 2015 for all the isolates tested in Bukavu and found an even distribution of the MIC values between the 2 years for ampicillin (Additional file 3). We concluded that the reproducibility was satisfactory for both the disc diffusion tests and the MIC determinations performed in Bukavu and all the results shown in the results section are therefore from the antibiotic susceptibility tests performed in Bukavu.

A multiplex real-time PCR, able to detect 40 different serotypes, was used according to a protocol published by Centers for Disease Control and Prevention (CDC) using previously published primers with slight modifications (preprint available at https://​www.​biorxiv.​org/​content/​early/​2018/​09/​12/​415422).

The multiplex real-time PCR assays were performed using the Quant Studio 6 Flex with a 384-well system (Applied Biosystems, Carlsbad, CA). Each reaction consisted of a 20 μL reaction volume, including 4 μL of template nucleic acid, along with 1 μM of each of the forward and reverse primers, 0.85 μM of the probe, 10 μl of 2X Universal Master Mix for DNA targets (Applied Biosystems) [31] and RNAase free water. The Tecan Freedom EVO PCR setup workstation (Tecan Group Ltd., Männedorf, Switzerland) was used to prepare the PCR reactions in a 384-well plate. The PCR reaction conditions were as follows: one initial cycle at 46 °C for 2 min, followed by denaturation at 95 °C for 10 min and 45 amplification cycles of 95 °C for 15 s and 58 °C for 1 min. Each multiplex performance was evaluated using an internal control (CpsA) to verify the presence of pneumococcal DNA in the sample, as well as two pUC57 plasmids containing each PCR target amplicon for all serotype systems.

For eight out of the 32 pneumococcal isolates that could be re-cultured after storage and transport to Gothenburg, Sweden, and in which the multiplex PCR serotyping method was inconclusive, the serotypes/serogroups were determined using a modified Sequetyping protocol (https://​www.​biorxiv.​org/​content/​early/​2018/​09/​12/​415422). Briefly, two PCR reactions were set up to amplified the whole cpsB gene. The PCR products were sent to GATC Biotech (Cologne, Germany) for purification and sequencing using the four PCR primers. The 1006 bp sequence product was matched to a reference database for determination of the serotype.

Descriptive analysis was performed using the SPSS package (version 24.0) for logistic regression to analyse the relationship between carriage and socio-demographic or medical factors. Prevalence rates and the 95% CI were calculated. Potential variables associated with pneumococcal carriage were assessed by odds ratios (OR) with 95% CI and tested by univariate analysis with the Pearson chi-square or Fisher’s exact test (n < 5). Associations with p < 0.05 were re-analysed by multivariate analysis. A p-value of < 0.05 was considered significant. Malnutrition was defined as the weight for age or weight for height as a Z score ≤ − 2 standard deviations, determined by ENA for smart software 2011.

From seven health care centres located in the city of Bukavu, in the suburban area or in the surrounding rural area, 794 children (age range one to 60 months, median 9.0 months) were included in the study and sampled from the nasopharynx. The background health data and living conditions of the children are shown in Additional file 1.

Overall, 163 (20.5%) of the children were culture positive for S. pneumoniae in the nasopharynx. The detection rate was associated with age but not with sex. Children aged 24–60 months had a more than three times higher rate of pneumococcal carriage that children below 6 months of age (p-value < 0.0001) (Table 2).

A higher frequency of pneumococcal carriage was observed in children living in the rural area as compared with the urban sites (28% versus 13%) and among children who lived in a house with an enclosed kitchen, i.e. with an open fire located inside the house, directly connected to the living room and/or the bedrooms, and these associations remained significant in multivariate analysis (Table 2). The type of stove and fuel for cooking did not correlate with carriage. Nor were there any associations between pneumococcal carriage and the number of rooms, having siblings, parents smoking tobacco, type of building material in the walls or the roof of the house, or having an animal in the household (Additional file 4).

Immunisation with PCV13 was strongly associated with lower rates of pneumococcal carriage, which was observed in only 3% of children who had received two or three doses of PCV13 as compared with approximately 30% of the unvaccinated children (p < 0.0001) (Table 2). Malnourished children, children with current fever and those who had had recent antibiotic treatment were more commonly colonised with pneumococci than children without these factors (p < 0.05). In contrast, neonatal problems, asthma, a recent history of malaria or gastroenteritis, or immunisation against measles, tuberculosis or Hemophilus influenzae type B were not associated with carriage, nor were symptoms of upper respiratory airway infection, such as a runny nose or cough (Table 2 and Additional file 5).

Taken together, age, living in a household with an enclosed kitchen, living in a rural area, undernutrition, current fever and antibiotic treatment during the last month were significantly associated with a higher, and vaccination with PCV13 with a lower, frequency of pneumococcal detection (Table 2).

The antimicrobial susceptibility pattern was determined at the Clinical Laboratory, Panzi Hospital, Bukavu, for the 163 pneumococcal strains that were isolated from the children (Fig. 2). Using disc diffusion tests, 145 (89%) of the isolates were shown to be non-susceptible to oxacillin and they were therefore regarded as resistant to phenoxymethylpenicillin. These 145 strains were further tested by MIC determination against penicillin G and 101 (62%) strains were categorised as intermediate (MIC 0.06–2 mg/L), while 30 (18%) were resistant (MIC > 2 mg/L). Taken together, 131/163 (80%) of the strains showed reduced susceptibility to benzylpenicillin, as confirmed by MIC determination (Fig. 2). Sixty-eight isolates (42%) had reduced susceptibility to ampicillin, of which 18 were resistant (MIC > 2 mg/L), and 61 isolates (37%) had reduced susceptibility to ceftriaxone (Fig. 2). High rates of non-susceptibility were also found for tetracycline and as many as 94% of the isolates were resistant to trimethoprim-sulphamethoxazole (TMP-SMX), also known as co-trimoxazole (Fig. 2). Notably, 70 (43%) of the pneumococci were multidrug resistant (non-susceptible to ≥3 classes of antimicrobial agents, including the beta-lactams).

The serotypes or serogroups of the isolated pneumococci were determined by multiplex real-time PCR or by the modified Sequetyping protocol, both performed in Gothenburg, Sweden. Among the 32 living isolates the most common serotype was 19F (n = 11), followed by 11A/D (n = 5) and 35B/35C (n = 5). The pneumococcal capsular cpsA gene was detected in all viable isolates, confirming their species identification.

The serotypes/serogroups of the pneumococci that could not be re-cultured were determined by multiplex PCR after isolation of genomic material from the non-viable isolates in the STGG storage medium (n = 119). In 62/119 cases (52%), one serotype/serogroup could be identified, whereas in 21 cases more than one serotype/serogroup was detected. Of these, two serotypes/groups were determined in 17 cases, three serotypes/groups in three cases, while one tube contained four serotypes/groups. In 36/119 cases (30%), no serotype or group could be determined by multiplex real-time PCR. The combined results for all 141serotypes/serogroups that were identified in the viable and non-viable pneumococcal isolates are shown in Fig. 3.

Hence, out of the 141 identified serotypes/serogroups, 76 (54%) belonged to a serotype/serogroup included in PCV13. However, these 76 included 13 pneumococcal strains that could not be distinguished between 6A, 6B, 6C and 6D, of which only 6A and 6B are included in PCV13. Two further strains could not be separated between 9A and 9 V, of which only 9 V is included. Sixty-five (46%) of the identified serotypes/serogroups could, however, be categorised as non-PCV13-containing types/groups. Thus, the proportion of identified serotypes/serogroups belonging to PCV13 was similar to those not belonging to the vaccine. In the nine children who had received two or three doses of PCV13, vaccine serotypes/groups (n = 9) were as commonly detected as were serotypes/groups not included in the vaccine (n = 7). There was no significant difference in the distribution of penicillin non-susceptibility between strains whose serotypes/serogroups are included in the PCV13 compared to strains with non-vaccine serotypes/groups (Additional file 6).