Background
Streptococcus pneumoniae (the pneumococcus) is among the top human pathogen with high rates of morbidity and mortality, with one of the largest public health concern and economic impact of any bacterial infectious agent in both developing and industrialized countries [
1‐
6]. It is reported that each year, approximately half of the 2.6 million deaths due to acute respiratory infections in under five year children are caused by pneumococcal pneumonia, the majority occurring in the developing countries [
3,
7]. It has also been estimated that in 2015 pneumococcal disease caused about 294, 000 deaths in children aged 1–59 months [
8].
Pneumococci are the major cause of common infections such as community-acquired pneumonia, otitis media, sinusitis, and also a major contributor to severe invasive infections such as septicaemia and meningitis. The nasopharynx of children is the ecological niche for pneumococci and nasopharyngeal colonization is a prerequisite for pneumococcal diseases. The polysaccharide capsule is a major pneumococcal virulence factor and more than 97 pneumococcal capsular serotypes have been identified with different potential to cause invasive pneumococcal disease (IPD) [
9‐
13]. The serotype distribution in pneumococcal diseases and carriage varies with time, geographical areas and age of the population. While some serotypes are found frequently in either or both disease and carriage, others are rarely isolated [
14]. Currently available pneumococcal vaccines are based on the pneumococcal capsule, either as a polysaccharide based vaccine targeting 23 serotypes (PPV23) or as conjugated vaccines targeting a limited number of serotypes, 7 in PCV7 (serotype 4, 6B, 9 V, 14, 18C, 19F and 23F), 10 in PCV10 PV7 plus serotypes 1, 5 and 7F) and 13 in PCV13 (PCV10 plus serotypes 3, 6A and 19A). Analysis of the serotype distribution in IPD and carriage as well as the disease incidence in a particular area is important to evaluate the effectiveness of the vaccines [
6,
15‐
17] .
A number of studies have shown that incidences of both invasive and non-invasive pneumococcal diseases have declined after PCV introduction in vaccinated children, and in some countries they also observed a herd protection effect in non-vaccinated populations such as the adults [
18‐
20] . However, other countries have not observed any herd protection effects on IPD in non-vaccinated elderly due to an almost complete replacement of vaccine types with non-vaccine types, both in IPD and in carriage in small children [
21,
22]. A reduction in pneumococcal carriage of PCV7 serotypes and an increase in colonization by non-PCV7 serotypes was observed after introduction of PCV7 in year 2000 in the United states [
23]. From year 2010, PCV7 was changed to either PCV10 or PCV13 in European countries and in the United States [
24‐
27] .
Ethiopia introduced PCV10 in the routine vaccination program in October 2011 to be given at the age of 6 weeks, 10 weeks and 14 weeks (3 + 0) without catch-up vaccination for older children. A few studies have reported nasopharyngeal carriage rates and found rates ranging from 41 to 78% [
28‐
30], but there is still a need for adequate baseline information on epidemiological factors such as the rate of carriage and transmission, serotypes and genetic relatedness of isolates, for subsequent impact assessment in Ethiopia. The objective of this study was to determine the carriage rates and analyze the phenotypic and genetic diversity of pneumococcal isolates in children before the first dose and after completion of PCV10 vaccination in the post-vaccination era.
Discussion
Nasopharyngeal carriage is a prerequisite for pneumococcal transmission and invasive disease. Pneumococcal colonization starts during early infancy and the median age at first acquisition of carriage was reported to be 33 days of age in The Gambia [
44], 38.5 days in Kenya [
45], 8 weeks in Bangladesh [
46] and 45.5 days in Thailand-Myanmar [
47]. In our study, we found a carriage rate of 26.6% at the age of 6 weeks when the infants came for the first dose of the PCV10 vaccine. This early exposure in infancy is likely to play an important role in the development of infection at an early age.
Nasopharyngeal carriage rates of pneumococci have been shown to vary depending on geographic region and population [
48]. In African and Asian countries, pneumococcal carriage rates were often reported to be high. In Africa, these included carriage rates of nearly 90% in The Gambia in children [
44], and other reports from the continent range from 18.6% to as high as 93.4% [
49‐
52]. Our finding of carriage rates, in fully vaccinated children at 9 months 56.8% and at 2 years 47.3%, is rather consistent with the high rate of carriage reported from most African countries and few studies done in Ethiopia [
28‐
30]. This suggests that vaccine has no impact on the overall carriage rate.
Carriage studies can be used to predict the potential impact and effectiveness of different formulations of PCVs [
53]. To our knowledge, this is the first study determining carriage rates of pneumococci and analyzing the impact of PCV10 immunization on pneumococcal carriage rates in Ethiopia using a longitudinal approach. Pneumococcal carriage rates increased significantly from 26.6% at the age of 6 weeks to 56.8% at the age of 9 months. Then carriage rates declined slightly to 48.3% when the children reached the age of 2 years. Similar situations have been reported from other studies in Malawi [
54], Kenya [
55] and The Gambia [
44]. A number of studies have shown that introduction of PCVs into routine vaccination programs did not affect overall pneumococcal carriage rates, but rather commonly led to non-vaccine type replacement [
25,
27,
55‐
57]. Hence, the overall carriage rate might not be affected by vaccination, while the serotype distribution might change, and this seems to have been the case in our study as well.
We identified 54 different serotypes among the at the age of 6 weeks isolates before the first dose of the vaccine, and about 80% of the isolates were of non-vaccine types, showing a highly diverse serotype distribution, which is in agreement with a report before vaccine introduction [
28]. This highlights the presence of a sizable pool of strains in Ethiopia with the potential for replacement of vaccine type strains with non-vaccine types in invasive and non-invasive diseases. This will probably have an important impact on the effectiveness of vaccination.
Out of 117 isolates from 9 months old children who completed the vaccination, we identified 43 different serotypes of which 89% were non-vaccine types. Vaccine types decreased from 20.2% before vaccination to 11.1% after use of the vaccine at the age of 9 months. Similarly, out of 97 isolates from children at 2 years, 90% were of non-vaccine types, and we did not find a reduction of vaccine types in this age group, when we compared to the age of 9 months. Our finding is in agreement with the Kenyan study where the carriage rate of PCV10 vaccine serotypes were significantly reduced from 34% before the immunization to 13% after completion of vaccination at two years [
58].
Several strains that belonged to PCV10 serotypes were detected in vaccinated children. These included 13 strains of serotypes 1, 4, 6B, 19F, 23F isolated from children at the age of nine months, and ten strains that belonged to serotypes 1, 4, 19F and 23F detected at the age of 2 years. The proportion of PCV10 serotypes such as 23F, 14 and 5 was higher before vaccination, and we did not detect any serotype 5, 14 and 9 V at the age of 9 months and 2 years. We did not identify any PCV10 serotypes 7F or 18C strains in our collection, but we detected related non-vaccine types 7C and 18F, serotypes that have been identified in clinical invasive specimens (CSF, blood, ear discharge, throat, pleural fluid, and sputum) from Gondar and Addis Ababa [
59,
60]. Why children continue to be colonized with vaccine types after completion of the vaccination schedule is unknown and could be due to several reasons. One possibility is that the concentration of antibodies produced as a result of carriage may not be sufficient to prevent colonization. Higher concentrations of IgG might be required to avoid colonisation than those necessary to prevent IPD [
61,
62].
Among the non-vaccine types found in our study, 13% were of serotypes 9 N, 20, 11A, 8, 16F, and these serotypes were previously reported to cause meningitis in children in Tikur Anbassa Hospital, Ethiopia [
60]. In another study in Ethiopia, serotypes 8, 10A, 13, 20, 22A, 27 and 15C were isolated from clinical samples [
59]. We detected all these serotypes in the nasopharynx of children except for serotype 22A. Serotype 10A was detected at a higher frequency than the other serotypes.
Among the infants that carried pneumococci at the age of 6 weeks (n = 210), 37.6% were still carriers at the age of 9 months. However, at 9 months (n = 117), only 4 children were still colonized with the same serotype as that at the age of 6 weeks. The serotypes were 6A in two children and 23F and 33B in the other children. PFGE analysis confirmed that these at the age of 9 months isolates were different from those obtained at 6 weeks of age. This finding indicates that there is a natural fluctuation in carriage rates of different serotypes rather than a persistence of the first colonizer. Presence of the same serotype at two different time points does not necessarily mean that the same strain will continue colonizing. Colonization by different members of the same serotype replacing each other over time is therefore possible.
The proportion and distribution of circulating serotypes not included in currently available PCVs, as well as their level of invasiveness, will be key determinants of the overall impact of the vaccine on disease. From countries that currently use PCV10 or PCV13, there are reports showing that non-vaccine serotypes such as 10A, 12F, 15A, 15B, 15C, 22F, 24F, 33F, 35B and 38 cause IPD [
22,
26,
63,
64]. Serotypes 12F, 22F, 24F and 33F have been identified also to have high invasive disease potentials [
26,
63‐
66]. All these non-vaccine types were identified in our study, and serotype 6A which is not included in PCV10, but part of PCV13, was a predominant serotype.
There are certain limitations with our study. The investigation was started one year and three months after the introduction of the vaccine and therefore, the first baseline data in our study might not be free of the effect of serotype replacement and herd protection. Also, only 116 children were sampled three times due to logistic problems. Moreover, due to high costs for MLST, we could only analyse 12 representative isolates from at the age of 6 weeks and 9 months isolates, selected based on the PFGE results. Furthermore, only a single colony was taken from each positive sample for serotyping, and hence we might have missed multiple serotypes in our analyses.