Potential of bacterial vaccines in the prevention of acute otitis media

Patients with acute otitis media (AOM) compose a large part of the daily practice of most general practitioners and pediatricians. It has been calculated that 0.5 million attacks of AOM occur annually in Finland, 1 and ∼24.5 million visits are made each year to US physicians for the treatment of OM. 2 The estimated annual costs caused by OM are US $138 million in Finland and $2 to 3.8 billion in the United States. 1, 3, 4 Prevention of even a small portion of the AOM cases would have a major impact from both the public health and economics points of view. In this article the published data on the role of the three most significant causative agents of AOM, Streptococcus pneumoniae, Haemophilus influenzae and Moraxella catarrhalis, are reviewed. Although other bacteria may also cause AOM, specifically Streptococcus pyogenes, Staphylococcus aureus and several Gram-negative bacteria, they are not discussed herein. This is part because of their minor and somewhat controversial role as true pathogens in AOM, but mostly because the development of vaccines has not progressed to the point where one might consider them viable candidates. Recent data from two Finnish prospective studies are presented to illustrate the potential impact of pneumococcal conjugate vaccines in the prevention of AOM. Additionally the development of other new vaccines based on antigens from S. pneumoniae, H. influenzae and M. catarrhalis are reviewed. These vaccines are still in the preclinical phase of development, so their eventual impact on AOM remains speculative. BACTERIAL ETIOLOGY OF AOM Our current understanding of the etiology of AOM is based on studies that all reported culture results from a considerable number of AOM events and middle ear fluid (MEF) samples, but which address different questions and had varying study designs. Diagnostic criteria of AOM, when defined, were variable from study to study, and more importantly, patients with protracted or complicated disease have often been overrepresented among the study subjects. Additionally the difference in the sensitivity of various bacteria to transportation and culture certainly has an impact on the reported percentages of various bacteria as causative organisms of AOM. Thus the figures cited within should be interpreted cautiously. There is, however, wide consensus that S. pneumoniae, H. influenzae and M. catarrhalis are the predominant bacteria associated with AOM. They can be found from MEF in ∼60% of the children with AOM (Table 1). Table 1: Percentage of bacterial pathogens in acute otitis media*S. pneumoniae has been reported most frequently as the bacterial pathogen causing AOM since the 1940s. 5, 6 Since then the number of AOM cases caused by S. pneumoniae has ranged from 18 to 52% in published reports, demonstrating a fluctuation, but not an increase or decrease in trend. 6, 7 The roles and importance of H. influenzae and M. catarrhalis as causative agents of AOM have only recently been fully appreciated. The proportion of MEF samples positive for H. influenzae has varied from 16 to 52%. 6, 7H. influenzae requires special growing conditions and has been associated with certain age and specific clinical characteristics. Most of the variation among study results may be the result of differences in the laboratory methods and in patient characteristics. The number of AOM cases caused by M. catarrhalis has increased during recent decades. In the 1970s and 1980s M. catarrhalis was cultured from 0 to 11% of MEF samples. Murphy 8 reviewed studies published in 1990 to 1994 and found that 11 to 23% of MEF samples were positive for M. catarrhalis by culture. A recent multinational treatment trial of AOM reported that of 837 bacterial isolates from MEF, 30% were S. pneumoniae, 27% were H. influenzae and 3% were M. catarrhalis. 9 Recent clinical data comes from the Finnish Otitis Media (FinOM) Cohort Study. 10 During 1994 to 1995, 329 children were enrolled in the trial, which studied the epidemiology and etiology of AOM and tested the procedures to be used in the efficacy trial of the new pneumococcal conjugate vaccine (FinOM Vaccine Trial). Children were enrolled at the age of 2 months and followed through the age of 24 months. In case of AOM with marked symptoms, myringotomy with aspiration was performed; MEF was collected for bacterial culture. Diagnosis of AOM was made 871 times by the study physicians and 1089 MEF samples were obtained from 772 AOM events (T Kilpi, M.D., unpublished data, 1999). S. pneumoniae was isolated in connection with 201 events (26%), H. influenzae in 174 events (23%) and M. catarrhalis in 177 events (23%). At least 2 of the 3 major pathogens were associated with 8% of the events. EFFICACY OF PNEUMOCOCCAL POLYSACCHARIDE AND CONJUGATE VACCINES The first bacterial vaccines tested for prevention of AOM in infants and children were the multivalent pneumococcal polysaccharide vaccines. Several studies were conducted in the 1970s for which the initial results were encouraging. 11–14 In the first study 827 Finnish otitis-prone children between the ages of 3 months and 6 years were enrolled in an efficacy study of the 14-valent pneumococcal polysaccharide vaccine. The attack rates of AOM (resulting from 100 vaccinees during a 6-month follow-up period) from 16.9 in the vaccinated group vs. 25.8 in the control group for those children <2 years of age; attack rates were 4.1 vs. 7.2 for the older children. 11 A subsequent study was conducted among a larger group of 3340 infants at the highest risk age (i.e. 7 to 9 months); however, no effect was demonstrated. During a 13-month follow-up, the attack rates per 100 vaccinated children were 16.0 and 15.3 in the vaccine and placebo groups, respectively. 15 One American study indicated that infants of African-American descent had fewer recurrences of AOM after vaccination than white infants. 13 On the contrary no effect in the rates of AOM was observed in children vaccinated in another American study or in an Australian study, when the vaccinated group was compared with the control group. 12, 14 Immunogenicity of polysaccharide antigens can be improved by coupling them with carrier protein. This method has been applied successfully to pneumococcal vaccines since the early 1990s. 16 New pneumococcal conjugates are immunogenic in infants, 17–23 able to induce immunologic memory, 18, 22, 24 produce antibody response on respiratory mucosa (A Nurkka, H Åhman, M Korkeila, V Jäntti, H Käyhty and J Eskola. Serum and salivary antibodies in infants and children immunized with the heptavalent pneumococcal conjugate vaccine. Submitted for publication) and reduce nasopharyngeal carriage of S. pneumoniae in immunized children. 25–28 Preliminary results from efficacy trials have recently become available. To date 2 trials have evaluated the efficacy of the new pneumococcal conjugate vaccines against AOM in infants. 29, 30 In the FinOM Vaccine Trial 1662 children were enrolled in a randomized, double blind efficacy trial of PncCRM, a 7-valent conjugate vaccine (pneumococcal serotypes 4, 6B, 9V, 14, 18C, 19F and 23F coupled with CRM197 protein). 29 Intramuscular injections of PncCRM or hepatitis B vaccine (i.e. control) were administered simultaneously with diphtheria-tetanus-pertussis/ H. influenzae type b (Hib) vaccine combination at 2, 4 and 6 months and with inactivated polio vaccine at 12 months of age. There were 2596 AOM episodes during the follow-up period for the per protocol analysis. One hundred seven AOM episodes in the PncCRM group and 250 episodes in the control group were caused by the pneumococcal serotypes in the vaccine. The reduction in the rate of number of episodes in the per protocol analysis was 57% against culture-confirmed, serotype-specific AOM, 34% against culture-confirmed pneumococcal AOM (irrespective of the serotype) and 6% against AOM irrespective of etiology (Table 2). 29Table 2: Preliminary results of the efficacy of the PncCRM vaccine in the prevention of acute otitis media29,31Another efficacy trial was conducted in northern California in 1995 to 1998. 30 In this trial 37 868 children were randomized to receive either PncCRM or meningococcal conjugate vaccine. The primary end point of the trial was the development of invasive pneumococcal disease. The investigators, however, were able to analyze the effectiveness of the vaccine on clinically diagnosed AOM using the records of the health maintenance organization. The reduction in AOM episodes (irrespective of the etiology) in the pneumococcal vaccine group during the per protocol follow period was 7%. 31 The overall impact of the PncCRM vaccine on AOM was very similar in both efficacy trials, suggesting that approximately one-third of the pneumococcal and ∼6 to 7% of all AOM events can be prevented with the PncCRM vaccine. OTHER EXPERIMENTAL PNEUMOCOCCAL VACCINES Although pneumococcal conjugate vaccines have a promising capability in the prevention of AOM, these vaccines may have certain shortcomings. Their efficacy, even against pneumococcal serotypes included in the vaccines, is <60%, and more effective vaccines would be highly desirable. 29 Recent articles have discussed innovative methods of how stronger immunity against pneumococcal polysaccharides may be induced in the future (Table 3). One method is to use pneumococcal vaccine with immunomodulating cytokines or with lymphocyte-stimulating antibodies. When pneumococcal polysaccharide is combined with pneumococcal surface protein A (PspA)/interleukin 2 fusion protein, vigorous antibody response against polysaccharide is seen in mice after subcutaneous immunization. 32 This mechanism appears to be related more to the immunomodulating properties of the cytokine molecule than to the carrier protein effect. Another method is to administer the vaccine together with CD40 monoclonal antibody. 33 This combination activates B cells without the direct help of T cells. Pneumococcal polysaccharide antigens have also been used as mucosal vaccines, together with novel adjuvants, and both serum and mucosal immune responses have been successfully induced. 34–37Table 3: Potential future pneumococcal vaccines and status of their developmentIt is not realistic to expect that the coverage of a conjugate vaccine would be >70% to 90% of the known pneumococcal serotypes causing OM. Thus some pneumococcal AOM would still occur even if vaccines were 100% efficacious. In this respect protein vaccines could have a definite advantage over polysaccharide-based vaccines, because the protection they may provide would not be restricted to the capsular serotypes included in the vaccine. In addition proteins can be expressed in recombinant bacteria at high levels, enabling large scale production of vaccine antigens at a relatively low cost. To be suitable for vaccine development, however, an antigen needs to be expressed in vivo, needs to be available to the host immune mechanisms and needs to be target of a protective immune response in human infants. Another requirement for a good vaccine candidate is that it be rather conserved, so that in the ideal case the immunity induced by the vaccine covers all the strains of the target pathogen. Inactivated pneumolysin toxin, PspA and a 37-kDa metal-binding lipoprotein (PsaA) are commonly mentioned as potential pneumococcal protein vaccines. 38 They can be used either as a pure protein, 39–41 conjugated with pneumococcal polysaccharides 42, 43 or as fusion proteins combined with immunomodulating molecules. 32 Mice immunized with inactivated or recombinant pneumolysin toxoid exhibited enhanced survival when challenged intraperitoneally or intranasally with several serotype strains of S. pneumoniae.40, 44 Similarly immunization with PspA or with truncated PspA elicits systemic and local immunity and protects mice against challenge with homologous or heterologous strains of S. pneumoniae.45–48 PsaA is a surface protein with significant homology with fimbrial adhesion proteins. Administered together with adjuvant it is able to protect mice against heterologous challenge. 41 All available evidence suggests that these three proteins all contribute to the virulence of S. pneumoniae, but they act at different stages of the pathogenic process. Thus there is a good possibility that a combination would provide a higher degree of protection than any one antigen alone. 49, 50 Even orally or intranasally administered protein antigens given with mucosal adjuvants induce protective antibody and cytokine responses in mice. 51–55 Other possible options to induce immune responses to S. pneumoniae are to deliver pneumococcal protein antigens either in recombinant carrier bacteria or in a form of nucleic acid vaccine. 39, 56 H. INFLUENZAE AND M. CATARRHALIS VACCINES Bactericidal antibody response and animal challenge studies have been used to identify protective antigens for future H. influenzae and M. catarrhalis vaccines (Table 4). These vaccines are still at an early stage of development. Even the structure and function of many of the molecules listed in Table 4 are still not well-understood. Another obstacle to a universal AOM vaccine is that there is yet to be any firm evidence that any single antibody to H. influenzae or M. catarrhalis are alone associated with protection against human AOM. Better animal models for studying H. influenzae and especially M. catarrhalis OM would also be desirable. Research in the area continues to be active and good progress has been made during recent years. Table 4: Potential antigens suggested for new acute otitis media vaccinesProtective antibodies have been generated by immunization of animals with various surface exposed antigens of H. influenzae, including purified outer membrane proteins (OMPs), 57–59 fimbriae (or peptides derived from them), 60 stress response proteins, 61 lipoprotein D (a surface-exposed immunoglobulin D-binding membrane protein of H. influenzae) 60, 62, 63 and lipooligosaccharides. 64 Immunity induced by several of these antigens has been protective in animal challenge studies, and antibodies induced by vaccination were protective in passive transfer experiments. Blocking attachment of bacteria to respiratory mucosa is a promising approach to prevent H. influenzae infection. The high molecular weight (HMW) adherence protein or fimbrial protein vaccines have been tested and proved protective in the chinchilla model. 63, 65 Also intranasal immunization with cell membrane preparation from H. influenzae or with OMP P6, together with mucosal adjuvants, seems to enhance clearance of H. influenzae from nasopharyngeal mucosa and inhibit its adherence to cultured human epithelial cells. 66–68 A major problem in the potential development of an H. influenzae vaccine is the heterogeneity of the surface structures between different strains. Some OMPs of H. influenzae (e.g. P6) show rather good homology, but surface determinants of others are highly variable, so that antibodies to surface epitopes of one strain will not bind to surface epitopes of another. 58 The problem of antigen variation seems to be especially prominent with fimbrial and lipooligosaccharide-based vaccines. When antibody responses to M. catarrhalis have been analyzed in children with AOM, several different antigens potentially useful in vaccine development (e.g. OMPs, lipooligosaccharides and fimbriae) have been identified. 69–72 The OMP profiles of different M. catarrhalis strains seem to be rather similar, and there are many cross-reacting epitopes. 73 Unlike H. influenzae OMP heterogeneity would therefore not appear to be a major barrier to the development of an OMP vaccine against M. catarrhalis. Among the most extensively studied molecules are HMW OMPs 70, 72, 74 and M. catarrhalis porin molecules (e.g. OMP CD and OMP E). 75–78 Two HMW OMPs, UspA1 and UspA2, contain both conserved and variable epitopes and are able to elicit biologically functional antibodies. 74, 79 Also OMP CD induces bactericidal antibodies, and immunization with it enhances pulmonary clearance of M. catarrhalis in a mouse model. 75, 76 It has proved immunogenic even through mucosal immunization. 76, 78 OMPs involved in iron acquisition (transferrin- and lactoferrin-binding proteins), as well as lipooligosaccharides also seem capable of providing protection against both homologous and heterologous strains of M. catarrhalis. 80–83 POTENTIAL OF BACTERIAL VACCINES IN THE FUTURE In the ideal world where vaccines of 100% efficacy against all strains of S. pneumoniae, H. influenzae and M. catarrhalis are developed, we could expect to prevent most although not all AOM cases in children. On the basis of the attack rates and the etiologic distribution in the FinOM Cohort Study, one could expect that in Finnish circumstances the ideal AOM vaccine would prevent nearly 200 of the 300 AOM episodes in the group of 100 children between the ages of 2 and 24 months. The world of vaccine development is, however, not ideal. We do not have H. influenzae, or M. catarrhalis vaccines, and the efficacy of the pneumococcal conjugate vaccine is not 100%. With the efficacy estimate obtained in the FinOM Vaccine Trial, 6% or 20 cases of those 300 episodes could be prevented with the PncCRM vaccine. Even though this figure is low, when compared with the efficacy estimates of other vaccines it is by no means negligible. In Finland the PncCRM vaccine could prevent ∼10 000 AOM cases annually among children at ages 2 to 24 months. Speculating that the efficacy of the vaccine would be 6% in all age groups, the PncCRM vaccine, if used widely, could prevent ∼30 000 cases of AOM in Finland and up to 1.5 million cases in the US annually. QUESTIONS/ANSWERS Question: What was the primary endpoint in the FinOM Vaccine Trial? Dr. Eskola: An episode of AOM due to pneumococcal serotypes in the 7-valent vaccine. A physical examination, including pneumatic otoscopy and tympanometry, was performed in a standardized way, and in case of AOM with marked symptoms, myringotomy with aspiration was performed and MEF collected for bacterial culture. The etiologic diagnosis was based on these culture findings and serotyping results. An episode was defined using an interval of 21 days (i.e. in cases when the same serotype was found within a period of 21 days from another MEF sample), it was considered to belong to the same episode. Question: Did you have an opportunity to look at the reduction in the number of office visits rather than episodes of OM? Dr. Eskola: Data on all visits are in the database and have yet to be fully analyzed. I must stress that this clinical study was not designed for health care utilization or health economics analyses. The families in the study were provided the best medical care we could offer free of charge, because we wanted to guarantee highest possible compliance. Parents were encouraged to bring their child for check-ups to their own physician in the study clinic whenever the child had symptoms of upper respiratory infection or if the parents suspected AOM for any reason. Therefore the number of sick visits may be higher than in situations outside the trial setting, and the results may not be applicable to the general population in this respect. Question: The data suggest that there was ∼50% prevention of OM cases caused by pneumococcal serotypes contained in the vaccine. Are there additional analyses on the patients who were not protected based on the relationship to onset after time of immunization, carriage or other factors that may explain the difference? Dr. Eskola: One of the important goals in our trial was to search for laboratory or epidemiologic surrogates for protection, so that we would in the future be able to predict the protection afforded by various vaccines without repeating full scale efficacy trials. Therefore, a large amount of information and samples had been collected during the FinOM Vaccine Trial. These will be analyzed both on population level as well as individually in the vaccine failure cases in the next few years.