Invasive Pneumococcal Disease in Children’s Hospitals: 2014–2017

The US Pediatric Multicenter Pneumococcal Surveillance Study Group consists of investigators from 8 children’s hospitals (Houston, TX; Pittsburgh, PA; Little Rock, AR; San Diego, CA; Los Angeles, CA; Chicago, IL; Columbus, OH; and Winston-Salem, NC) who have been prospectively identifying children with IPD since September 1993.¹⁵  We identified infants and children with IPD from our database seen between January 1, 2014, and December 31, 2017. IPD was documented by positive culture results from a normally sterile site (blood, cerebrospinal fluid, pleural fluid, synovial fluid, peritoneal fluid, etc). The diagnosis of pneumococcal pneumonia required an abnormal chest radiograph result plus a positive culture result from blood, pleural fluid, or lung. Mastoiditis was also considered an invasive infection for which isolates were obtained from middle ear fluid, subperiosteal abscess, or mastoid bone. The approach to obtaining blood cultures in children with fever without a source was inconsistent among the 8 children’s hospitals. Demographic and clinical information was collected retrospectively and recorded on a standard case report form for each episode of infection.

Administration of PCV7 or PCV13 was documented through the patient’s medical records by contacting the patient’s health care provider or in a vaccine registry. Invasive disease due to a vaccine serotype isolate occurring after receipt of PCV7 or PCV13 was counted only if the infection occurred at least 2 weeks after receipt of a PCV. A supplemental case report form for any immune evaluation testing was completed for patients who had received ≥2 PCV doses before their IPD episode if due to a PCV13 serotype isolate. The study has been reviewed by the institutional review boards of each of the participating hospitals.

The patient database was maintained in a central office (Texas Children’s Hospital and Baylor College of Medicine, Houston, TX). Pneumococcal isolates were identified by using standard methods in the microbiology laboratories of each hospital and sent to a central laboratory for further testing (Dr Edward O. Mason, Jr, Infectious Disease Research Laboratory, Texas Children’s Hospital, Houston, TX). Isolates were serotyped by the capsular swelling method by using commercially available antisera (Statens Serum Institut, Copenhagen, Denmark; Cedarlane Laboratories, Inc, Burlington, NC).¹⁵ 

Minimal inhibitory concentrations (MICs) for penicillin and ceftriaxone were performed and interpreted by standard methods.¹⁶  For noncentral nervous system infections, isolates were considered susceptible to parenteral penicillin if MICs were ≤2.0 µg/mL or to ceftriaxone if MICs were ≤1.0 µg/mL. For central nervous system infections, isolates were considered susceptible to parenteral penicillin if MICs were ≤0.06 µg/mL or to ceftriaxone if MICs were ≤0.5 µg/mL. Susceptibility data were not available for all isolates.

Statistical analysis was performed by using Fisher exact or Wilcoxon rank sum tests (Stata11; College Station, TX).

A total of 495 patients were identified over the years 2014–2017; 13 patients had no isolate and were excluded from the study (in Supplemental Fig 3, we show the annual number of IPD cases). Age distribution did not differ between patients with IPD due to PCV13 serotype versus non-PCV13 serotype isolates (Supplemental Fig 4). The total number of PCV13 and most common non-PCV13 serotype isolates are shown in Table 1. Overall, PCV13 serotype isolates accounted for 23.9% (115 out of 482) and non-PCV13 serotype isolates for 76.1% (367 out of 482) of IPD isolates over the 4 years of this study. PCV13 serotype isolates accounted for 20% (26 out of 128), 27% (29 out of 108), 25% (31 out of 124), and 24% (29 out of 122) of IPD cases in 2014, 2015, 2016, and 2017, respectively. The most common PCV13 serotypes were 3, 19A, and 19F, which together accounted for 91% of these isolates.

The most common non-PCV13 serotypes were 35B, 23B, 33F, and 22F, all of which combined accounted for 41% (149 out of 367) of these isolates. Serogroup 15 (serotypes 15A, 15B, and 15C) in aggregate accounted for 13% (49 out of 367) of the isolates. The sites of IPD caused by PCV13 serotype versus non-PCV13 serotype isolates are shown in Table 2. The distribution of sites of infection differed significantly (P < .0001) between the 2 groups. Pneumonia was the most common site of infection for PCV13 serotype isolates, whereas bacteremia was the most common site for non-PCV13 serotype isolates.

An underlying condition was significantly (P < .0001) more common among patients with IPD due to non-PCV13 serotype isolates (200 out of 367, 54.5%) compared with patients with IPD caused by PCV13 serotype isolates (27 out of 115, 23.5%) (Table 3). Leukemia and other malignancies were the most common underlying conditions for patients with non-PCV13 serotype isolates. Only non-PCV13 serotype isolates were recovered from the 17 children with hemoglobinopathies or asplenia. Children with underlying conditions were significantly older than patients without underlying conditions (P < .001) (Supplemental Fig 5).

The distribution of PCV13 serotypes and the number of PCV13 or PCV7 doses before IPD are shown in Table 4. (Data on vaccine status were not available for 10 children.) For the 105 children for whom PCV status was known, 30 had not received a PCV and 14 had received 1 dose before their IPD. Of the 30 children who had not received a PCV, 22 (73%) were of an age at the time of IPD for which ≥2 doses of PCV were recommended. Two, 3, or 4 doses were received by 6, 18, and 37 children, respectively. Twenty-four children had an underlying condition. Children with serotype 3 isolates causing their IPD were significantly less likely (P < .0001) to have an underlying condition (1 out of 45) than children with infections due to serotype 19A (10 out of 28) or serotype 19F (9 out of 24).

Twenty-two children with underlying conditions received PPSV23 before their IPD episode. Six had IPD due to a PPSV23 but non-PCV13 serotype organism (one each of serotypes 8, 10A, 15B, 15C [considered B/C], 22F, 33F). Five had IPD due to a 23A serotype organism, and 3 had IPD due to a 15B serotype organism. None had a PCV13 serotype isolate.

An immune evaluation of some type was documented in 28 of 61 children who received at least 2 doses of PCV13 and whose IPD was caused by a PCV13 serotype isolate (or PCV7 if a PCV7 serotype isolate caused their IPD) (Table 5). Of these 28 patients, one had a slightly low immunoglobulin M (40 mg/dL; normal for age: 44–155 mg/dL); a child with pre–B cell acute lymphocytic leukemia had a low immunoglobulin G (IgG) (377 mg/dL; normal for age: 423–1090 mg/dL) and low antibody count to the causative serotype (19F, 0.4 µg/mL with a level >15 µg/mL, considered protective by the laboratory performing the test). Another child with acute lymphocytic leukemia had low immunoglobulin A (22 mg/dL; normal for age: 54–150 mg/dL) and had a low antibody count to the causative serotype (19F, 0.26 µg/mL), and 1 had no protective antibody to tetanus toxin on a humoral immune panel. One patient who had received 4 doses of PCV13 was found to have a diagnosis of common variable immunodeficiency (CVID) established as a result of evaluating quantitative immunoglobulins (IgG: 332 mg/dL [normal for age: 546–1533 mg/dL]; immunoglobulin M: 53 mg/dL [normal for age: 21–95 mg/dL]; immunoglobulin A: 53 mg/dL [normal for age: 54–150 mg/dL]) after having pneumococcal meningitis due to a serotype 23F isolate at 3 years of age. However, this patient had a family history of CVID in a maternal grandfather, aunt, and cousin. One patient had abnormal T-cell testing, but no specific immunodeficiency disorder was assigned (testing was done on day 4 of IPD). Fifteen children had antibody measured to the serotypes (serotype 3, n = 9; 19F, n = 5; 19A, n = 1) causing their infection. All but 3 (2 patients with leukemia, another with a renal transplant) had concentrations much above those considered protective or normal by the laboratory performing the test.

The distributions of MICs for penicillin and ceftriaxone for PCV13 serotype and non-PCV13 serotype isolates are shown in Figs 1 and 2. The proportion of isolates with penicillin MIC ≤2.0 µg/mL was significantly greater (P < .0001) for non-PCV13 serotype isolates (339 out of 340) compared with PCV13 serotype isolates (85 out of 104). Similarly, the proportion of isolates with ceftriaxone MIC ≤1.0 µg/mL was significantly greater (P < .0001) for non-PCV13 serotype isolates (334 out of 334) compared with PCV13 serotype isolates (93 out of 99). Serotype 19A isolates accounted for all but 2 of the isolates with penicillin MICs >2.0 µg/mL. All but 6 isolates (all serotype 19A) in both groups combined had ceftriaxone MICs ≤1.0 µg/mL. For all isolates, 8.3% (36 out of 433) would be considered nonsusceptible to ceftriaxone for treating a central nervous system infection (17% [17 out of 99] for PCV13 serotype isolates versus 5.7% [19 out of 334] for non-PCV13 serotype isolates, P < .001).

In the United States, IPD in children has been markedly reduced because of the routine use of PCV7 beginning in 2000 followed by PCV13 in 2010. Most of the remaining cases are due to non-PCV13 serotypes. In 8 children’s hospitals in the United States, we found that 4 to 7 years after PCV13 was introduced, ∼25% of IPD cases in children were due to PCV13 serotype isolates, most commonly serotypes 3, 19A, and 19F. However, >40% of the children in our study who had IPD due to a PCV13 serotype isolate had received none or only 1 dose of PCV13 (or PCV7 if serotypes 14, 18C, 19F, or 23F) before their IPD.

Serotype 3 was the most common PCV13 serotype causing IPD in this study, but 42% (19 out of 45) of these patients had received none or only 1 dose of PCV13 before their IPD episode. In a study conducted by the Centers for Disease Control and Prevention and a recent systematic review and meta-analysis, it has been shown that the effectiveness of PCV13 is less for preventing IPD due to serotype 3 compared with serotypes 19A and 7F.^17,18  It has also been noted in other studies that PCV13 is less effective in preventing IPD due to serotype 3 compared with other PCV13 serotypes.^5,14,19  It is notable that only 2 of the patients in our study with IPD due to a serotype 3 isolate had an underlying condition. Serotype 19A was the second most common PCV13 serotype in this study, but >40% (12 out of 28) of these patients had received none or only 1 PCV13 dose. Others have reported serotype 19F among the more common PCV7 or PCV13 serotypes causing IPD or nasopharyngeal colonization in appropriately immunized children.^20–22  None of the 4 patients with IPD due to serotype 7F had received a dose of PCV13.

The most common non-PCV13 serotypes noted in our study were serotypes 35B, 23B, 33F, and 22F. Underlying conditions were noted in >50% of patients with non-PCV13 serotype IPD, which was significantly greater than the 23% of patients with underlying conditions and IPD caused by PCV13 serotype isolates. This may be related to the relatively decreased invasive potential of the non-PCV13 serotypes to cause IPD compared with PCV13 serotype isolates.²³  In the Active Bacterial Surveillance of IPD in the United States, the Centers for Disease Control and Prevention found that for children <5 years of age, the most common non-PCV13 serotypes for 2015–2016 were 23B, 22F, 33F, 15C, 15A, 35B, and 10A.²⁴  In Massachusetts children <18 years old, the most common non-PCV13 serotype isolates from April 2010 through March 2017 were 15BC, 33F, and 22F; PCV13 serotype isolates accounted for 33% of cases.²⁵  Other non-PCV13 serotype isolates are more common in reports from outside the United States; in addition, the relative frequency of the specific most common non-PCV13 serotypes from studies outside the United States differs from that noted in US studies.^26–28 

Although PCV13 serotype isolates were significantly more likely to be nonsusceptible to penicillin or ceftriaxone than non-PCV serotype isolates on the basis of the 2018 Clinical and Laboratory Standards Institute nonmeningeal site of infection interpretive breakpoints, still ∼95% and 98% of all isolates were susceptible to penicillin or ceftriaxone, respectively. However, almost 10% of all isolates had ceftriaxone MICs considered nonsusceptible for treating central nervous system infections, which supports continuing to recommend the combination of ceftriaxone plus vancomycin for empirical treatment of suspected pneumococcal meningitis.²⁹ 

Gaschignard et al³⁰  conducted a prospective study of children with IPD from 2005 to 2011 in 28 pediatric units in France. Of 163 patients (142 with meningitis) identified, 127 had complete and 35 had incomplete immunologic evaluations. Previous IPD had occurred in 17 patients, and 4 other patients had a history of other previous severe infections. Abnormal immunologic responses were found in 26 children (16%). Primary immune deficiencies were found more commonly in children >2 years of age (14 out of 53, 26%) compared with younger children (3 out of 109, 3%). The authors advocate for a systematic immunologic exploration for all children hospitalized for IPD. In a study from the United Kingdom, serum from 172 children with IPD submitted to the Health Protection Agency for measuring serotype-specific IgG antibody concentrations also had total serum IgG concentrations measured.³¹  Nineteen (11%) had low IgG concentrations, although the levels were only marginally below the lower limit of normal for age. Three of the 19 children had a repeat IgG measurement 3 to 5 months later, and all 3 had IgG levels in the normal range, suggesting a transient IgG deficiency.

The 2015 American Academy of Pediatrics Red Book states that a child’s HIV status and immunologic function should be considered for evaluation if they have IPD due to an isolate with a serotype in PCV13 and have received ≥2 doses of PCV13 at least 2 weeks before the onset of illness.³²  In our study, 28 children with IPD due to a PCV13 serotype isolate and who had received at least 2 doses of PCV13 (or PCV7 if a PCV7 serotype isolate caused IPD) had a full or partial immune evaluation at the discretion of their treating physicians. Only 1 patient was found to have a true immunodeficiency, and that patient had a strong family history of CVID. It is of interest that this is the only patient who had meningitis despite 4 doses of PCV13. In our study, it is suggested that in an otherwise healthy child who has received ≥2 doses of PCV13 and subsequently develops IPD due to a PCV13 serotype isolate but without other evidence of an immunodeficiency from the history, physical examination, or routine laboratory studies, an immune evaluation is unlikely to be revealing. However, further studies are needed to determine the value of an immune evaluation for such patients.

Our study has several limitations. Not all information is available for all patients, including PCV administration. Some isolates were not available for testing. The serotypes of the isolates were not known in real time for most patients, and immune evaluations were done in only half of the children for whom this would be a consideration on the basis of the 2015 Red Book statement. Furthermore, the immune evaluations were not standardized, and many were composed primarily of quantitative immunoglobulin testing.

PCV13 serotypes (especially serotypes 3, 19A, and 19F) were responsible for 25% of IPD among children at 8 children’s hospitals in the United States 4 to 7 years after the routine administration of PCV13 to infants was introduced. Almost 30% had not received any doses of PCV13 before the IPD episode, although >70% of these children were at an age at the time of the IPD at which they should have received ≥2 doses. Thus, additional cases of IPD might be prevented if PCV13 and PPSV23 are administered to all eligible children according to the recommended schedule. Considering >50% of the patients with non-PCV13 serotypes causing IPD had an underlying condition that could also be complicated by poor immune response to vaccination, further decreasing IPD in this group of children may be challenging. Finally, an immune evaluation for children who have IPD due to a PCV13 serotype isolate despite receiving ≥2 doses of PCV13 is unlikely to be helpful, unless other clinical information suggests an immunodeficiency.

CVID

common variable immunodeficiency

IgG

immunoglobulin G

IPD

invasive pneumococcal disease

MIC

minimal inhibitory concentration

PCV

pneumococcal conjugate vaccine

PCV7

7-valent pneumococcal conjugate vaccine

PCV13

13-valent pneumococcal conjugate vaccine

PPSV23

23-valent pneumococcal polysaccharide vaccine