Pneumococci Isolated From Children in Community-Based Practice Differ From Isolates Identified by Population- and Laboratory-Based Invasive Disease Surveillance

Abstract

Background

Characterizing strains causing noninvasive and invasive pneumococcal disease (IPD) may inform the impact of new pneumococcal conjugate vaccines (PCVs).

Methods

During 2011–2019, among children aged 6–36 months, pneumococcal serotype distribution and antibiotic nonsusceptibility of nasopharyngeal and middle ear fluid (MEF) isolates collected at onset of acute otitis media (AOM) in Rochester, New York, were compared with IPD isolates from the Active Bacterial Core surveillance (ABCs) system across 10 US sites.

Results

From Rochester, 400 (nasopharyngeal) and 156 (MEF) pneumococcal isolates were collected from 259 children. From ABCs, 907 sterile-site isolates were collected from 896 children. Non-PCV serotypes 35B and 21 were more frequent among the Rochester AOM cases, while serotypes 3, 19A, 22F, 33F, 10A, and 12F contained in PCVs were more frequent among ABCs IPD cases. The proportion of antibiotic-nonsusceptible pneumococcal isolates was generally more common among IPD cases. In 2015–2019, serotype 35B emerged as the most common serotype associated with multiclass antibiotic nonsusceptibility for both the Rochester AOM and ABCs IPD cases.

Conclusions

Pneumococcal isolates from children in Rochester with AOM differ in serotype distribution and antibiotic susceptibility compared to IPD cases identified through US surveillance. Non-PCV serotype 35B emerged as a common cause of AOM and IPD.

Pneumococcal isolates from nasopharyngeal colonization and middle ear fluid from community-based children with acute otitis media in Rochester, New York, differ in serotype distribution and antibiotic susceptibility compared to invasive pneumococcal disease isolates from multisite active laboratory- and population-based surveillance.

Higher-valency pneumococcal conjugate vaccines (PCVs) that include 15 serotypes (PCV15) or 20 serotypes (PCV20) have recently been licensed in the United States (US) and other countries [1]. Even higher-valency pneumococcal vaccines are in advanced stages of development [2, 3]. While PCV15 and PCV20 include common serotypes associated with disease, their effectiveness in preventing nasopharyngeal colonization, noninvasive mucosal infections such as acute otitis media (AOM), acute sinusitis, and nonbacteremic community-acquired pneumonia (CAP) and invasive pneumococcal disease (IPD) has yet to be understood as these vaccines were licensed based on safety and immunogenicity data [1, 4].

IPD has high morbidity and mortality but is uncommon, whereas noninvasive mucosal pneumococcal infections are more common, especially AOM in young children [5]. Prevention of nasopharyngeal colonization and noninvasive mucosal pneumococcal infections requires higher levels of serum antibody than the levels required to prevent IPD [6, 7]. Mucosal antibody levels [8], T resident memory cells [9], and T-helper 17 cells [9–11] likely also play an important role. Lower levels of serum antibody following immunization with higher-valency PCVs compared with 13-valent PCV (PCV13) for some included serotypes might be anticipated due to antigenic competition and has been reported for PCV20 [12].

Divergence of serotypes that are prevalent in causing less common but potentially fatal IPD versus more common mucosal infections has occurred [13–17]. In addition, vaccine pressure and acquisition of antibiotic resistance may contribute to PCV serotype replacement with non-PCV serotypes [15, 18], reducing the net effectiveness of deployed PCVs, as was observed globally post–PCV7 introduction with serotype 19A [19]. Therefore, knowledge of serotypes and antibiotic resistance of prevalent non-vaccine-serotype pneumococcal strains colonizing the nasopharynx and comparison with isolates causing disease help identify pneumococcal strains that are likely to emerge as pathogenic.

In this study, we describe serotype distributions and antibiotic susceptibility patterns of pneumococci from 2 populations of children aged 6–36 months: (1) strains isolated from the nasopharynx or middle ear fluid (MEF) at onset of AOM of children seen in community-based pediatric practices in Rochester, New York (NY); (2) strains causing IPD among children from the Centers for Disease Control and Prevention's (CDC) Active Bacterial Core surveillance (ABCs), a laboratory- and population-based surveillance network across 10 US sites.

METHODS

The years of comparison were 2011–2019, after PCV13 introduction to the US national immunization program, up to the beginning of the coronavirus disease 2019 pandemic.

Study Populations

Rochester AOM Cases

Pneumococcal isolates collected from a healthy child population in Rochester, NY, prospectively enrolled at age 6 months and followed until age 36 months were included in the analyses. Details of this cohort have been previously described [13, 20, 21]. In brief, at onset of clinically diagnosed AOM, nasopharyngeal (NP) cultures were obtained to detect and characterize pneumococci [22]. AOM clinical diagnostic criteria recommended by the American Academy of Pediatrics were used and tympanocentesis was performed for first and any subsequent AOM episodes to collect MEF for detection and characterization of pneumococci. All enrolled children had received a full primary series of 3 doses of PCV13 (given at 2, 4, and 6 months of age) and a booster dose at 15 months of age. The study was approved by the Rochester Regional Health Institutional Review Board and registered at ClinicalTrials.gov (NCT04946084).

ABCs IPD Cases

The ABCs catchment population includes California (3 counties); Colorado (5 counties); Connecticut; Georgia (20 counties); Maryland (6 counties); Minnesota; New Mexico; New York (16 counties including Rochester area); Oregon (3 counties); and Tennessee (20 counties). IPD cases among children aged 6–36 months in the ABCs catchment area with pneumococcal isolates were included in this analysis. A case of IPD was defined as isolation or detection of Streptococcus pneumoniae from a normally sterile site, such as blood, cerebrospinal fluid, or pleural fluid [23]. For each case, medical records were reviewed to obtain demographic and clinical information. IPD cases with AOM were defined as a clinical diagnosis of AOM or mastoiditis noted in the medical chart (not based on collection of MEF using tympanocentesis), along with S pneumoniae isolation from a normally sterile body site. The ABCs protocol was reviewed in accordance with the CDC human research protection procedures and was determined to be nonresearch, public health surveillance. Participating sites also obtained any required approvals from their state, local, and hospital institutional review boards.

Microbiology and Serotyping

Samples from the Rochester AOM cases were tested by standard microbiology techniques to identify the presence of pneumococci as previously described [14]. Pneumococcal capsular serotypes were determined by Quellung reaction (Statens Serum Institut, Denmark) on pure cultures of pneumococci.

ABCs laboratory methods, which include Quellung reaction, polymerase chain reaction, or whole-genome sequencing (WGS), have been previously described [24]. Pneumococcal serotypes were grouped into PCV13 (1, 3, 4, 5, 6A, 6B, 7F, 9V, 14, 18C, 19A, 19F, 23F), PCV15 non-PCV13 (22F, 33F), PCV20 non-PCV15 (8, 10A, 11A, 12F, 15B), and non-PCV20 (all other serotypes) types.

Antibiotic Susceptibility Testing

From the Rochester AOM cases, pneumococcal isolate susceptibility to 16 different antibiotics was determined with the VITEK-2 automated modified broth microdilution system using the gram-positive susceptibility card AST-GP68 (discontinued) and AST-GP74 (bioMérieux, Durham, North Carolina) in the clinical laboratories of Rochester General Hospital, Rochester, NY. Benzylpenicillin, amoxicillin, ceftriaxone, cefotaxime, meropenem, ertapenem, ofloxacin, levofloxacin, moxifloxacin, erythromycin, telithromycin, vancomycin, linezolid, tetracycline, chloramphenicol, and trimethoprim-sulfamethoxazole (TMP-SMX) susceptibility were tested.

For the ABCs cases, minimum inhibitory concentration (MIC) by broth microdilution was determined as previously described [25] for pneumococcal isolates from 2011–2015. For pneumococcal isolates from 2015–2019, predictions were inferred from WGS data [26], including the penicillin binding protein typing system for determining β-lactam antibiotic MICs [27]. Conventional MIC testing of selected strains was also performed. The different methodologies provide similar results for interpretation [26].

Pneumococci were classified as susceptible, intermediate, or resistant based on 2017 Clinical and Laboratory Standards Institute breakpoints [28]. Multiclass nonsusceptibility was defined as nonsusceptibility to ≥3 classes of antibiotics [24] (Supplementary Materials).

Analysis Groups and Statistics

Pneumococci from the Rochester AOM cases were compared with pneumococci from 2 groups of ABCs IPD cases: (1) cases from all ABCs sites, and (2) cases from the NY ABCs site, which includes Rochester, to account for potential geographic differences in circulating strains. Differences in demographic characteristics, serotype distribution, and antibiotic susceptibility pattern among Rochester AOM and ABCs IPD cases were assessed individually by 2-sided Fisher exact test, with P < .05 considered significant. Analysis of serotype distribution and antibiotic susceptibility was based on pneumococcal isolates, rather than children, as we considered pneumococcal isolations as independent events for comparisons between the NP isolates at onset of AOM and IPD isolates and similarly between the MEF isolates. Statistical analyses were conducted using GraphPad Prism software, version 6.

RESULTS

Rochester AOM cases were reported in 259 children: 253 children with pneumococcal colonization in the nasopharynx at the time of AOM diagnosis (NP population) and 126 children with pneumococci isolated from MEF (MEF population); 120 children were included in both NP and MEF populations. The ABCs IPD cases included 896 children from all ABCs sites and 84 children from the NY site. Compared with Rochester AOM cases, IPD cases occurred in children with older median age (7 vs 16–17.5 months) and a larger proportion of Black race (4%–4.7% vs 32.8%–33.3%) (Table 1). Children with underlying conditions were excluded per protocol in Rochester, whereas 9% of children with IPD had an underlying condition that was an indication for pneumococcal vaccination [29].

Table 1.

Demographic Characteristics of Study Populations of Children Aged 6–36 Months

Characteristic	Rochester, NY NP Colonization at AOM	Rochester, NY Middle Ear Fluid at AOM	ABCs All Sites Invasive Cases	ABCs NY Site Invasive Cases
	(n = 253)a	(n = 126)b	(n = 896)c	(n = 84)d
Sexe
Male	155 (61.3)	78 (61.9)	528 (58.9)	48 (57.1)
Female	98 (38.7)	48 (38.1)	368 (41.1)	36 (42.9)
Age, mo, median (IQR)	7 (6–11)	7 (6–11)	16 (11–24)	17.5 (12–25)
Racef
White	201 (79.4)	107 (84.9)	472 (52.7)	49 (58.3)
Black	12 (4.7)	5 (4)	294 (32.8)	28 (33.3)
American Indian/Alaska Native	0	0	35 (3.9)	0
Asian/Pacific Islander	4 (1.6)	0	53 (5.9)	5 (6.0)
≥2/Other	19 (7.5)	9 (7.1)	5 (0.6)	0
Unknown	17 (6.7)	5 (4)	37 (4.1)	2 (2.4)
Ethnicitye
Hispanic	25 (9.9)	10 (7.9)	110 (12.2)	7 (8.3)
Non-Hispanic	228 (90.1)	116 (92.1)	735 (82.0)	74 (88.1)
Unknown	0	0	51 (5.7)	3 (3.6)

Data are presented as No. (%) unless otherwise indicated.

Abbreviations: ABCs, Active Bacterial Core surveillance; AOM, acute otitis media; IQR, interquartile range; NP, nasopharyngeal; NY, New York.

^aThese 253 children had 400 pneumococcal isolates in the nasopharynx; 168 children had 1 isolate, while 47 children had 2 isolates, 21 children had 3 isolates, 12 children had 4 isolates, and 5 children had 5 isolates, each sampled at different AOM episodes. One child had 2 distinct pneumococcal isolates sampled in the nasopharynx during 1 AOM episode. Of the children with pneumococcal colonization in the nasopharynx, 120 also had pneumococcal detection in the MEF and are included in the “Rochester, NY Middle Ear Fluid at AOM” column.

^bThese 126 children had 156 pneumococcal isolates in the MEF; 104 children had 1 isolate, 14 children had 2 isolates, and 6 children had 3 isolates, each sampled at different AOM cases. Three children had 2 distinct pneumococcal isolates sampled in the MEF during 1 AOM episode, each.

^cEight hundred ninety-six children had 907 IPD isolates; 11 children had 2 isolates from separate IPD episodes.

^dEighty-four children had 86 IPD isolates; 2 children had 2 isolates from separate IPD episodes.

^eNo statistically significant difference between the populations for sex or ethnicity.

^fHigher proportion of White children in Rochester NP and MEF populations compared to ABCs all sites and ABCs NY sites (P ≤ .0001 for every comparison), and lower proportion of Black children in Rochester NP and MEF populations compared to ABCs all sites and ABCs NY sites (P ≤ .0001 for every comparison).

Clinical Syndromes Associated With Pneumococcal Isolates

Among the 259 children in Rochester with AOM, no IPD cases occurred. Among the NP population, 400 pneumococcal isolates were identified during 399 AOM episodes. Among the MEF population, 156 pneumococcal isolates were detected during 153 episodes of AOM (Table 2). In 4 children, >1 isolate was detected simultaneously colonizing the nasopharynx or in the MEF.

Table 2.

Frequency of Pneumococcal Isolation of Study Populations by Clinical Syndrome

Clinical Syndromea	Rochester, NY AOM Cases With NP Colonization	Rochester, NY AOM Cases With MEF Infection	ABCs All Sites Invasive Cases	ABCs NY Invasive Cases
	(n = 399)	(n = 153)	(n = 907)d	(n = 86)e
AOM	399 (100.0)b	153 (100.0)c	98 (10.8)	9 (10.5)
Bacteremia without focus	0	0	414 (45.6)	44 (51)
Bacteremic pneumonia	0	0	259 (28.6)	22 (25.6)
Meningitis	0	0	87 (9.6)	6 (7.0)

Data are presented as No. (%).

Abbreviations: ABCs, Active Bacterial Core surveillance; AOM, acute otitis media; MEF, middle ear fluid; NP, nasopharyngeal; NY, New York.

^aNot mutually exclusive.

^bFour hundred NP isolates at onset of AOM from 399 episodes of 253 children; some children had multiple AOM episodes. In 1 case, 2 pneumococcal isolates were found colonizing the nasopharynx simultaneously.

^cOne hundred fifty-six MEF isolates detected in the MEF of 153 AOM episodes from 126 children; some children had multiple AOM episodes. In 3 cases, 2 pneumococcal isolates were detected simultaneously in the MEF during a single AOM episode.

^dEight hundred ninety-six children had 907 IPD isolates; 11 children had 2 isolates from separate IPD episodes.

^eEighty-four children had 86 IPD isolates; 2 children had 2 isolates from separate IPD episodes.

Among all ABCs sites, 907 IPD cases (11 children had 2 separate IPD episodes) were reported. Bacteremia without focus was the most commonly associated syndrome (45.8%), followed by bacteremic pneumonia (28.6%) (Table 2). The distribution of reported clinical syndromes in a subset of IPD cases from the NY ABCs site was similar to that of all ABCs sites.

Pneumococcal Serotypes

Serotypes 19A, 19F, and 3 were the most common PCV13 serotypes in all 3 groups, and the proportion of serotypes 19A and 3 was significantly higher among IPD isolates compared with the NP isolates (Figure 1). The proportion of PCV15 non-PCV13 serotypes (22F, 33F) was significantly higher in the IPD isolates compared with both NP and MEF isolates. Among PCV20 non-PCV15 serotypes, the proportion of serotype 10A was significantly higher among IPD isolates compared with NP isolates, whereas the proportion of serotypes 11A and 15B was significantly higher among the NP and MEF isolates compared with the IPD isolates. Among the non-PCV20 serotypes, serotypes 35B, 23B, 21, 15A, and 23A were among the most common serotypes for the NP and MEF isolates, and the proportions were significantly higher compared with the IPD isolates. On the other hand, the proportion of serotype 38 was significantly higher among the IPD isolates compared with the NP or MEF isolates. Differences in serotype distribution between IPD and NP or MEF isolates remained even after limiting the IPD cases to those from the NY site (Supplementary Table 1). Within Rochester AOM cases, pneumococcal serotype distribution among NP (n = 400) and MEF (n = 156) isolates was not significantly different (Figure 1). Differences in serotype distribution between Black and White children were small, although the number of pneumococcal isolates from Black children in Rochester was limited (Supplementary Results, Supplementary Table 2). By study period, the most common serotypes among the AOM NP (n = 158) isolates in 2011–2014 were 15B (15%), 35B (11%), 23B (8%), 21 (8%), and 15A (8%), whereas in 2015–2019 (n = 242), serotypes 35B (19%), 23B (12%), 21 (10%), 15A (7%), and 15B (6%) were the most common. Among ABCs cases in 2011–2014 (n = 495), serotypes 33F (14%), 22F (12%), 19A (10%), 15C (8%), and 38 (6%) were most common, and in 2015–2019 (n = 412), serotypes 33F (11%), 22F (9%), 15C (8%), 23B (7%), and 35B (7%) were most common.

Figure 1.

Pneumococcal serotypes of Rochester, New York vs Active Bacterial Core surveillance all sites of children aged 6–36 months, 2011–2019. There was no statistically significant difference between the Rochester nasopharyngeal (NP) and middle ear fluid (MEF) populations. The 400 NP isolates came from 253 children, while the 156 MEF isolates came from 126 children; 120 children are present in both groups. Eight hundred ninety-six children had 907 invasive pneumococcal disease (IPD) isolates; 11 children had 2 isolates from separate IPD episodes. *Statistically significant difference for NP vs IPD isolates only. ^$Statistically significant difference for both NP vs IPD and MEF vs IPD isolates. ^†Non-PCV20 serotypes that represent <1% in any group. ^¶Isolates that were not fully typed because they are uncommon (such as 25/38/43/44/45/46/48) or isolates not fully subtyped (such as 24A/24B/24F). Unknown isolates were not used in determining statistical significance. Abbreviations: ABCs, Active Bacterial Core surveillance; AOM, acute otitis media; IPD, invasive pneumococcal disease; MEF, middle ear fluid; NP, nasopharyngeal; PCV, pneumococcal conjugate vaccine.

Pneumococcal Antibiotic Susceptibility

Antibiotic susceptibility testing was completed on 224 (56%) NP and 85 (55%) MEF pneumococcal isolates from Rochester AOM cases and 899 (99%) from ABCs IPD isolates (Table 3). The proportion of nonsusceptible isolates was the highest for erythromycin (41%–48%), oral penicillin (27%–33%), and TMP-SMX (20%–34%) across all 3 groups. Compared with NP isolates, IPD isolates had a significantly higher proportion nonsusceptible to TMP-SMX (34% vs 20%, P < .0001), tetracycline (16% vs 7%, P = .0006), ceftriaxone (4% vs 1%, nonmeningitis breakpoints, P = .04), cefotaxime (5% vs 1%, nonmeningitis breakpoints, P = .03), and chloramphenicol (2% vs 0%, P = .02), whereas the proportion nonsusceptible to meropenem was lower among IPD specimens compared with NP (18% vs 12%, P = .02). Compared with MEF isolates, IPD isolates had a significantly higher proportion nonsusceptible to cefotaxime (5% vs 0%, nonmeningitis breakpoints, P = .04), ceftriaxone (13% vs 5%, meningitis breakpoints, P = .02), and TMP-SMX (34% vs 20%, P = .008). Similar antibiotic susceptibility patterns were observed when limiting the ABCs cases to the NY ABCs site only. Multiclass nonsusceptible pneumococcal isolates were identified in 46 of 309 (15%) of the Rochester AOM isolates and in 208 of 900 (23%) ABCs isolates. In both AOM and ABCs isolates, the most common serotypes with multiclass nonsusceptibility in 2011–2014 were serotypes 19A (50% and 33%, respectively) and 35B (30% and 20%, respectively). In 2015–2019, serotype 35B was the most common in both AOM and ABCs isolates (64% and 27%, respectively), followed by serotypes 15A/B/C (AOM isolates: 15B [14%], 15C [8%], and 15A [8%]; ABCs isolates: 15A [18%], 15C [13%], 15B [11%]). Comparisons of serotype distributions and antibiotic susceptibility patterns between Rochester AOM isolates and a subset of ABCs isolates associated with an AOM diagnosis were similar to those of the main analysis (Supplementary Tables 3–5, Supplementary Figure 1).

Table 3.

Pneumococcal Antibiotic Susceptibility by Study Population

Antibiotic Tested	Antibiotic Susceptibility Range			R	I	S	% of Non-susceptible (I + R)	R	I	S	% of Non-susceptible (I + R)	R	I	S	% of Non-susceptible (I + R)
				Rochester, NY AOM NPa				Rochester, NY AOM MEFb				ABCs All Sites Invasive Cases
	R	I	S	(n = 224)c				(n = 85)c				(n = 899)d,e
Penicillin (oral)	≥2	0.12 ≤ I ≤ 1	≤0.06	8	66	148	33%	7	16	61	27%	109	148	642	29%
Penicillin (parenteral, nonmeningitis)	≥8	4	≤2	1	5	216	3%	1	6	77	8%	16	31	852	5%
Amoxicillin	≥8	4	≤2	2	14	205	7%	1	3	77	5%	44	55	797	11%
Cefotaxime (meningitis)	≥2	1	≤0.5	3	24	197	12%	0	11	74	13%	40	66	790	12%
Cefotaxime (nonmeningitis)	≥4	2	≤1	1	2	221	1%	0	0	85	0%	13	27	856	5%
Ceftriaxone (meningitis)	≥2	1	≤0.5	3	6	215	4%	1	3	81	5%	38	78	783	13%
Ceftriaxone (nonmeningitis)	≥4	2	≤1	1	2	221	1%	0	1	84	1%	13	25	861	4%
Meropenem	≥1	0.5	≤0.25	16	25	182	18%	5	10	69	18%	58	53	788	12%
Levofloxacin	≥8	4	≤2	0	0	224	0%	0	0	85	0%	0	0	899	0%
Erythromycin	≥1	0.5	≤0.25	77	1	84	48%	23	0	31	43%	367	0	532	41%
Linezolid	…	…	≤2	0	0	224	0%	0	0	85	0%	0	0	899	0%
Vancomycin	…	…	≤1	0	0	224	0%	0	0	85	0%	0	0	899	0%
Tetracycline	≥8	4	≤2	14	2	208	7%	7	0	78	8%	142	1	756	16%
Chloramphenicol	≥8	…	≤4	0	0	222	0%	0	0	84	0%	19	0	880	2%
TMP-SMX	≥4	1.2	≤0.5	22	22	180	20%	9	8	68	20%	95	213	591	34%

Numbers in bold denote statistically significant differences in the proportion of antibiotic-nonsusceptible pneumococcal isolates when compared with ABCs invasive pneumococcal disease (IPD) isolates.

Abbreviations: ABCs, Active Bacterial Core surveillance; AOM, acute otitis media; I, intermediate; MEF, middle ear fluid; NP, nasopharyngeal; NY, New York; R, resistant; S, sensitive; TMP-SMX, trimethoprim-sulfamethoxazole.

^aThe proportion of isolates nonsusceptible to meropenem was higher in the Rochester NP population compared to the ABCs all sites population (P = .0215). The proportion nonsusceptible to cefotaxime (nonmeningitis; P = .0308), ceftriaxone (meningitis, P ≤ .0001 and nonmeningitis, P = .0444), tetracycline (P = .0005), chloramphenicol (P = .0204), and TMP-SMX (P ≤ .0001) was lower in the Rochester NP population compared to the ABCs all sites population. There was no statistically significant difference between the Rochester NP and MEF populations.

^bCefotaxime (nonmeningitis; P = .0421), ceftriaxone (meningitis; P = .0238), and TMP-SMX (P = .0077) nonsusceptibility is lower in the Rochester MEF population compared to the ABCs all sites population.

^cThe 224 NP isolates came from 156 children, while the 85 MEF isolates came from 69 children; 51 children are present in both groups.

^dEight hundred eighty-eight children had 899 IPD isolates with susceptibility data available; 11 children had 2 isolates from separate IPD episodes.

^eSusceptibility data for amoxicillin and cefotaxime (meningitis and nonmeningitis) was only available for 896 of the 899 IPD isolates.

DISCUSSION

Our study compared pneumococcal strains from a cohort of children aged 6–36 months in Rochester presenting with AOM (without IPD) and children with IPD identified from population- and laboratory-based surveillance during 2011–2019, after PCV13 introduction. Non-PCV serotypes 35B and 21 were significantly more frequent from the AOM cases, while serotypes 3, 19A, 22F, 33F, 10A, and 12F contained in 13-valent, 15-valent, or 20-valent PCVs were significantly more frequent from the ABCs cases. In general, there was a larger proportion of nonsusceptible pneumococcal isolates from the ABCs cases. In 2015–2019, serotype 35B emerged as the most common serotype associated with multiclass antibiotic nonsusceptibility for both the AOM and ABCs isolates.

While prior studies have compared differences in pneumococcal strains identified from NP carriage and IPD [30, 31], there are very limited published data that directly compared strains from AOM and IPD. Nonvaccine types have emerged to cause disease, vary by country, vary by adult versus pediatric populations, and are dynamically changing year to year [32, 33]. Pneumococci expressing certain serotypes are more invasive than others [34] and the distribution of serotypes differs between different disease states [13, 35], contributing to the differences in serotype distribution between AOM and IPD cases. While differences in geographic distribution and demographics of the 2 populations in our study may be contributing to some of the observed differences, a high proportion of serotypes 35B and 15B among children with AOM has been seen in other US settings. Among pneumococcal isolates identified from MEF in children aged ≤5 years with complicated otitis media through a hospital-based antimicrobial surveillance program, 35B and 15B were the most frequently identified pneumococcal serotypes [36]. At the same time, compared with the Rochester cohort in our study, these children with complicated otitis media had a higher proportion of PCV13 serotypes, primarily serotypes 19A, 19F, and 3 (MEF isolates from hospital-based surveillance vs Rochester cohort: 22.6% vs 10.2%), and the proportion of non-PCV20 serotypes was lower (47.2% vs 58.3%), similar to what we observed in the ABCs IPD isolates. This may be because isolates identified through hospital-based surveillance represent a more severe presentation of AOM cases [30], and serotypes 3 and 19A are considered to have higher invasive capacity compared with other pneumococcal serotypes. Additionally, the hospital-based complicated otitis media is expected to be inherently different from a well-established community-based population regarding vaccination coverage and prior antibiotic exposure, increasing the risk of disease due to serotypes that are nonsusceptible to antibiotics such as serotypes 19A and 15B.

Increase in serotype 35B in pneumococcal carriage, AOM, and IPD cases in the US has been previously reported. NP colonization of serotype 35B increased among children in Rochester after PCV13 introduction in children in 2010; however, the proportion of AOM cases did not increase until 2015, and since then, it has been the most predominant serotype identified among pneumococcal AOM cases in Rochester [18]. In another carriage study conducted among children in Georgia, increase in the proportion of serotype 35B carriage was seen earlier, after PCV7 introduction and before PCV13 introduction [37]. In both studies, ST558 was the predominant serotype 35B strain, contributing to a large proportion of penicillin-nonsusceptible pneumococcal isolates. In IPD cases, the increase in serotype 35B cases has been primarily in older adults aged ≥65 years (ABCs unpublished data), and most of these isolates were ST558 of penicillin nonsusceptible clonal complex 558.

Analysis of ABCs IPD cases of all ages during 2014–2018 showed that serotype 35B is now the main cause of penicillin-nonsusceptible IPD cases, followed by serotype 19A [24, 38]. In addition to the high carriage rate in children and the high proportion of antibiotic nonsusceptibility, clonal expansion and serotype switching of serotype 35B strains have been observed, which are features similar to serotype 19A after PCV7 introduction. Serotype replacement in IPD cases has not been reported in the US post–PCV13 introduction, but close monitoring of pneumococcal disease trends is warranted as the proportion of isolates covered by new higher-valency PCVs (eg, PCV15, PCV20) that were recently recommended in the US do not include many of the serotypes that were commonly seen among AOM cases, including serotype 35B.

While this study focuses on pediatric disease and suggests that the serotype distribution and antibiotic susceptibility differ between AOM and invasive disease, serotype distribution and antibiotic susceptibility may differ in adults relating to another mucosal infection, namely nonbacteremic CAP. Indeed, serotype-specific urine antigen detection assays suggest that the serotype distribution of adult pneumococcal pneumonia may differ from that of AOM in children [39, 40].

Our analysis is subject to several limitations. First, differences in children with AOM in Rochester and IPD in ABCs could have contributed to some of the observed differences between the 2 populations. There were notable differences in the racial distribution among the Rochester AOM and ABCs populations, although we did not see substantial differences in serotype distribution by race of children. Children in the Rochester cohort all had received PCV13 doses appropriate for age and children with underlying conditions were excluded, whereas previous analysis of ABCs data showed that approximately 5% of children with IPD (17% if limited to cases with PCV13-type IPD) had never received a dose of PCV and had a larger proportion of underlying conditions compared with community controls matched by age and zip code [41]. Thus, differences in PCV13 coverage and underlying health conditions (therefore, likelihood of prior antibiotic exposure) could be contributing to the higher proportion of PCV13 serotypes observed among IPD cases. Second, nasopharyngeal cultures taken at onset of AOM identify pneumococci that might cause AOM, but such samples are not consistently concordant with MEF cultures [42]. Third, given that sequence typing was not performed consistently across the study period for both the Rochester and ABCs isolates, we were unable to compare the genomic characteristics of the pneumococcal isolates from the 2 populations.

Despite these limitations, our study provides important insight into differences in pneumococcal strains identified from 2 unique data sources that provide comprehensive assessments of children with AOM in Rochester, NY, and children with IPD from ABCs (including Rochester) that may have future policy implications. The impact of the newly recommended PCV15 and PCV20 against pneumococcal disease burden and emergence of replacement serotypes are yet to be seen, and further divergence in predominant pneumococcal serotypes causing mucosal and invasive disease may occur. New, higher-valency pneumococcal vaccines are already in advanced stages of development. An investigational 21-valent PCV includes 8 additional serotypes that are not included in PCV15 or PCV20, serotypes 15A, 15C, 16F, 23A, 23B, 24F, 31, and 35B [2]. This 21-valent vaccine does not include serotypes 1, 2, 4, 5, 6B, 9V, 14, 15B, 18C, 19F, and 23F that are contained in currently available pneumococcal vaccines targeted primarily for adults. Other higher-valency PCVs are in development for adult and/or pediatric populations that expand coverage beyond that of PCV15 and PCV20 or complement coverage achieved by PCV15 and PCV20. As IPD becomes more uncommon among US children, the importance of demonstrated efficacy of new vaccines against noninvasive mucosal infections, such as AOM, sinusitis, and nonbacteremic pneumonia, may become increasingly important. Continued monitoring of circulating strains causing noninvasive and invasive pneumococcal disease in the US will likely provide insight to optimize future pneumococcal vaccine policies.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.

Notes

Author contributions. R.K., R.G., M.K. and MP. participated in conception and design of the study; analysis and interpretation; drafting and critical review of the final manuscript. L.M. and E.G. participated in analyss and interpretation and critical review of the final manuscript. ABCs team participated in acquistion and critical review of the final manuscript. All authors had full access to all the data and take full responsibility for the manuscript findings.

Acknowledgments. Investigators and affiliations of the Active Bacterial Core surveillance team: Arthur Reingold (University of California, Berkeley); Maria Rosales (California Emerging Infections Program, Oakland); Meghan Barnes (Colorado Department of Public Health and Environment, Denver); Susan Petit (Connecticut Department of Public Health, Hartford); Monica M. Farley (Emory University and Atlanta Veterans Affairs Medical Center, Atlanta, Georgia); Lee H. Harrison (Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland); Ruth Lynfield and Corinne Holtzman (Minnesota Department of Health, St Paul); Kathy M. Angeles, Sabra Arias, Jessica Houston, Sarah A. Khanlian, Mayvilynne Poblete, and Zachary Q. Landis (New Mexico Emerging Infections Program, Santa Fe); Rachel Wester, Kari Burzlaff, Bridget J. Anderson, Suzanne McGuire, and Jemma V. Rowlands (New York State Department of Health, Albany); Ann Thomas and Tasha Martin (Oregon Public Health Division, Portland); and William Schaffner, H. Keipp Talbot, and Tiffanie M. Markus (Vanderbilt University, Nashville, Tennessee). We also acknowledge Janet Casey, MD (Legacy Pediatrics), Steven Schulz, MD (Finger Lakes Medical Associates, Pediatrics) and Andrew Sherman, MD (Bay Creek Pediatrics), who provided clinical leadership and secured clinical samples. Jasmine Mathoan and Olivia Riggs provided technical assistance. Roberto Vargas, PhD, Director of the Rochester General Hospital clinical laboratory, provided oversight of antibiotic susceptibility testing. We thank everyone in the ABCs areas who are involved in surveillance and maintenance of the system at the 10 sites. We also thank the laboratorians and technicians who isolate the ABCs pathogens and make it possible to track these infections, as well as the surveillance and laboratory personnel at the CDC for their careful work characterizing the isolates. We acknowledge Mirasol Apostol, Kathryn Como-Sabetti, Lori Triden, Paula Snippes, Anita Glennen, and Kerry MacInnes of the ABCs team and others for their contributions at the study sites, and Tamara Pilishvili for her leadership on this project.

Disclaimer. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the CDC.

Financial support. The funding source for the collection of pneumococcal isolates from the Rochester cohort was the US National Institute on Deafness and Communication Disorders (grant number R0108671, principal investigator [PI]: M. P.) and CDC (contract number 75D30119C06842, PI: M. P.). For isolates from the ABCs, the funding source was the CDC. Support for the analysis and preparation of this manuscript was from the CDC (PI: M. P., 75D30121C12195).