Streptococcus pneumoniae (pneumococcus) remains one of the most important causes of morbidity and mortality in adults and children throughout the world.[1] It is estimated that this microorganism is responsible for more than 100 000 000 cases of ear infections in children, 5 000 000 cases of pneumonia and 100 000 cases of meningitis — the whole being associated with 10 000 000 cases of bacteraemia every year.[1] Mortality still remains unacceptably high, despite all advances in medicine, including the availability of potent antimicrobial therapy, improved medical and nursing care, and even the establishment of intensive care unit (ICU) facilities. Furthermore, management is potentially complicated by the emergence of resistance in the pneumococcus to the commonly used antibacterials. This review focuses on the appropriate management of patients with pneumococcal bacteraemia, with particular reference to the antibacterial management of patients with community-acquired pneumonia (CAP).

1. Pneumococcal Infection

1.1 Epidemiology, Risk Factors and Prognosis

Interest in pneumococcal infection remains high, which is not surprising considering that this pathogen is the most common bacterial cause of CAP, meningitis and otitis media. A number of recent reviews have highlighted the ongoing impact of pneumococcal infections, particularly in the setting of CAP, and especially in association with bacteraemia.[2–4] The incidence of pneumococcal bacteraemia has been estimated to be 5.8/100 000 inhabitants/year, although a downward trend has been identified recently, most likely as a consequence of effective pneumococcal vaccination.[4] Of the estimated 5 000 000 cases of pneumococcal pneumonia occurring in the US per year, bacteraemia is present in approximately 10–20% of patients, and the mortality remains high even in patients treated appropriately with antibacterials (10–255).[4] Almost 10% of infections are complicated by septic metastases to distant organs, causing complications such as meningitis, endocarditis, empyema, peritonitis and various others.[4]

The incidence of invasive pneumococcal disease (IPD) varies substantially and is affected by factors such as socioeconomic status, age, immune status, genetic background and geographical location.[3] Certainly, there are a number of well defined risk factors for pneumococcal infection and bacteraemia in both adults and children.[3] For example, in one study cigarette smoking was found to be one of the strongest independent risk factors for IPD in immunocompetent, non-elderly adults.[5] Even more recently, a comparative study of cases with bacteraemic and non-bacteraemic pneumococcal pneumonia indicated that smoking was the leading risk factor for pneumonia, and while current smokers had an increased risk of bacteraemia, former smokers and chronic obstructive pulmonary disease (COPD) patients developed non-bacteraemia forms more commonly.[6] Interestingly, although there is some debate about the issue, at least some studies indicate that the outcomes of bacteraemic pneumococcal pneumonia in COPD patients are better than expected, with mortality lower in COPD than in non-COPD patients.[7]

Similarly, HIV infection is a considerable risk factor for pneumococcal infections and especially bacteraemic infections, although trends in hospitalizations for IPD appear to be decreasing in countries such as the US since the introduction of pneumococcal conjugate vaccination in children.[8] Some studies have suggested that there are few differences in the presentation of bacteraemic and non-bacteraemic pneumococcal pneumonia.[9] However, a more recent study comparing HIV-infected and non-infected patients with bacteraemic pneumococcal pneumonia indicated that when adjustments were made for age and severity of illness, HIV-infected cases had a significantly higher 14-day mortality, with a trend to increasing mortality with lower CD4+ cell counts.[10] Similarly, a study of CAP occurring in HIV-infected patients indicated that in those patients who were not on antiretroviral therapy who had positive S. pneumoniae antigenuria, there was an increased risk of bacteraemia, and that bacteraemic patients had a poorer outcome.[11]

A number of studies have addressed the question of poor prognostic factors in pneumococcal bacteraemia.[12,13] In addition to those factors already described, older age, greater extent of pulmonary consolidation, need for mechanical ventilation/ICU admission and specific pneumococcal serotypes were associated with a worse outcome.[12,13] The case fatality rate for bacteraemic pneumococcal CAP varies in different parts of the world, being 20% in the US and Spain, 13% in the UK, 8% in Sweden and 6% in Canada, according to one study.[12] Differences in the severity of the disease at presentation, as well as presence and impact of underlying chronic conditions, most likely accounted for these differences. Certainly, pneumococcal pneumonia continues to be associated with considerable costs worldwide.[14]

1.2 Pathogenesis of Pneumococcal Infections

1.2.1 Colonization

Nasopharyngeal colonization of the non-immune host precedes, and is a prerequisite for, the development of invasive disease.[15] Successful colonization necessitates adherence of the pneumococcus to respiratory epithelium, an event that can only be realised if the pathogen survives its early encounters with the innate defence mechanisms of the respiratory tract, most importantly the expulsive actions of the mucociliary escalator. Subversion of the mucociliary escalator is achieved through the coordinated action of an array of protein and non-protein virulence factors. Foremost amongst these is the polysaccharide capsule, which, in addition to promoting resistance to opsonophagocytosis and detection by pattern recognition receptors, enables the pneumococcus to evade entrapment by mucopolysaccharides present in respiratory tract mucus.[16] Several additional virulence factors, most notably hydrogen peroxide (H2O2), pneumolysin and hyaluronidase, act directly on respiratory epithelium, causing ciliary slowing and epithelial damage.

Toxins

Through the action of a membrane-bound pyruvate oxidase, the pneumococcus produces prodigious quantities of H2O2, reaching low millimolar concentrations in bacteriological culture medium.[17] H2O2 is an indiscriminate, cell-permeable, reactive oxidant, which is toxic for both eukaryotic and prokaryotic cells. Several mechanisms may protect the catalase-negative pneumococcus against the autotoxic actions of H2O2. These include exclusion of oxidation-sensitive cysteine residues from exported and cytosolic proteins,[18] as well as a possible barrier and/or oxidant-scavenging function of the polysaccharide capsule.[19] Importantly, however, H2O2 is cytotoxic for ciliated respiratory epithelium, causing dysfunction of the mucociliary escalator.[20] Pneumolysin, which is usually released upon autolysis of the pneumococcus, is considered to be a key protein virulence factor of the pathogen.[16,21] It is a 53 kDa protein, which belongs to the family of cholesterol-binding, pore-forming, cytolytic microbial toxins. Like H2O2, pneumolysin has potent inhibitory effects on the integrity of ciliated respiratory epithelium, causing both ciliary slowing and epithelial damage.[22] These detrimental effects of pneumolysin, and possibly those of H2O2, are augmented by pneumococcal hyaluronidase, an enzyme that disrupts intercellular adhesion, thereby increasing the exposure of ciliated respiratory epithelium to both cytotoxins.[23] The pneumococcus therefore utilizes a seemingly unique combination of virulence factors, namely the polysaccharide capsule, H2O2, pneumolysin and hyaluronidase, to inactivate the mucociliary escalator, enabling the pathogen to adhere to the respiratory epithelium.

Adhesins

Attachment of the pneumococcus to the epithelium involves an array of bacterial adhesins. In the initial stages, adhesion is likely to be mediated predominantly by the non-protein virulence factor, phosphorylcholine, which interacts with the platelet-activating factor (PAF) receptor on the epithelium.[24,25] The C-polysaccharide of the pneumococcus, as well as some types of capsular polysaccharide, contains phosphorylcholine.[26,27] Phosphorylcholine/PAF receptor-mediated adhesion of the pneumococcus is reinforced by various protein adhesins, including the pneumococcal surface proteins (Psp) A and C (also known as choline-binding protein A [CbpA]) and the lipoprotein pneumococcal surface adhesin (Psa) A, which interact with the epithelial polymeric immunoglobulin (Ig) receptor that normally transports secretory IgA and E-cadherin, the cell-cell junction protein of respiratory epithelium, respectively.[24,28] Some serotypes also possess pilus-like structures that promote epithelial adhesion via interaction with uncharacterized receptors.[29] A novel protein adhesin has been described recently. This is the 120 kDa plasminogen- and fibronectin-binding protein B (Pfb B), which significantly increases the ability of the pneumococcus to adhere to epithelial cells.[30] Unmasking of these various pneumococcal protein adhesins necessitates a reduction in capsule size, with the accompanying risk of increased vulnerability to phagocytosis. This risk is apparently minimized by production of biofilm, a process in which bacterial neuraminidase plays an important role.[31,32] The pneumococcus expresses up to three cell surface neuraminidases (Nan A, B, C), which cleave terminal sialic acids from glycan chains on host cells,[18] exposing potential binding sites for bacterial adhesins, as well as inducing biofilm formation by free sialic acid.[32]

Once colonization is established, several virulence mechanisms are utilized by the pneumococcus to repel innate and adaptive host defences. These include (i) enzymatic modification of cell-wall peptidoglycans, rendering them resistant to lysozyme present in respiratory secretions;[33] (ii) cleavage of secretory IgA by a zinc metalloproteinase;[24,25] (iii) interference with activation of the alternative and classical complement pathways by PspA/PspC and pneumolysin, respectively;[24,34] and (iv) encasement in biofilm as described below.[35]

Biofilm Formation

Biofilm plays an important role in microbial colonization and persistence. It is a hydrated, self-generated polymer matrix in which microbial pathogens are effectively insulated, not only against the cellular and humoral defence mechanisms of the host but also against antibacterials (recently reviewed by Hall-Stoodley and Stoodley[35]). Concealed in biofilm, either on the epithelial surface or sequestered intracellularly,[35] the pneumococcus can re-emerge at times when host defences are compromised, as may occur during infection with influenza virus, respiratory syncytial virus and HIV-1, resulting in invasive disease.[36–38]

1.2.2 Invasive Disease

As described in a recent review, the progression from colonization of the nasopharynx to invasive infection is likely to involve a complex interplay between the virulence of the infecting strain of the pneumococcus and the efficiency of anti-pneumococcal host defences.[2] The transition from the relatively innocuous carrier state in the nasopharynx to being a dangerous, invasive pathogen appears to coincide with reversion to higher levels of capsule expression,[18] possibly by quorum-sensing mechanisms. PspA/PspC-mediated transcytosis of the pathogen across the epithelial barrier via the polymeric Ig receptor enables direct access of the pathogen to the bloodstream and invasion of the CNS.[24,25] Spread to the lungs, on the other hand, is most likely to occur by aspiration, with the probability of active infection heightened by a preceding respiratory virus infection.[36,38] Influenza virus infection results in prolonged exposure of pulmonary macrophages to interferon-γ, resulting in decreased phagocytic activity of these cells and increased susceptibility to pneumococcal infection.[39] Notwithstanding the involvement of the capsule, pneumolysin is a critical virulence determinant in the pathogenesis of pneumococcal pneumonia.[22]

Pneumolysin

In addition to its cytolytic activity, pneumolysin, at sub-lytic concentrations, possesses a range of potentially harmful, proinflammatory activities, primarily affecting epithelial cells and cells of the innate immune system, especially neutrophils and monocytes/macrophages (reviewed by Feldman and Anderson[2]). These result both from the pore-forming activities of the toxin, leading to influx of extracellular calcium, as well as from its interactions with Toll-like receptor (TLR)-4.[2,40–44] Activation of intracellular signalling pathways, including those involving p38 and JNK (c-Jun N-terminal kinase) mitogen-activated protein kinases, as well as nuclear factor (NF)-κB, leads to the production of proinflammatory cytokines/chemokines including interleukin (IL)-8, monocyte chemotactic protein 1 and tumour necrosis factor (TNF). In the case of neutrophils, interactions of these cells with sub-lethal concentrations of pneumolysin results in exaggerated release of reactive oxygen species, granule proteases and leukotriene (LT) B4.[45,46] Evidence in support of the involvement of pneumolysin in the pathogenesis of severe pneumococcal pneumonia has largely been derived from murine models of experimental infection. Feldman and colleagues[22] reported that injection of recombinant pneumolysin into the apical lobe bronchus of rats resulted in the development of a severe lobar pneumonia restricted to the apical lobe. More recently, Witzenrath et al.[47] and Garcia-Suárez et al.[48] have provided additional interesting insights into the role of pneumolysin in the pathogenesis of acute lung injury (recently reviewed by Feldman and Anderson[2]).

Witzenrath et al.[47] demonstrated that delivery of recombinant pneumolysin into the airways of mice resulted in increased capillary permeability and severe lung oedema, while intravascular administration of the toxin was accompanied by increased pulmonary vascular resistance and lung microvascular permeability. These authors concluded that pneumolysin may play a central role in early-onset acute lung injury by causing impairment of pulmonary microvascular barrier function and severe pulmonary hypertension.[47] They attributed these effects of the toxin, all of which are important features of acute respiratory distress syndrome, to its direct cytotoxic actions on pulmonary endothelial and epithelial cells, as opposed to proinflammatory activities.

Using a murine model of experimental pneumonia in which the mice were infected intranasally with the pneumococcus,[48] Garcia-Suárez et al.[48] reported that pneumolysin was detected in the lungs at sub-lytic concentrations and was located in epithelial cells, macrophages and leukocytes, but not vascular endothelial cells. They concluded that the proinflammatory activity of pneumolysin was the major factor in causing tissue damage in their model of pneumococcal pneumonia.

Taken together, the findings of these various studies suggest that in severe pneumococcal pneumonia, it is the combined effects of the cytotoxic and proinflammatory activities of pneumolysin that lead to acute lung injury and respiratory failure, as well as the epithelial damage that results in translocation of pneumococci from the alveoli to the interstitium and then the bloodstream.

1.3 Anti-Pneumococcal Host Defences

Anti-pneumococcal host defences have recently been reviewed elsewhere.[2] With respect to innate immunity, the following mechanisms initiate a predominantly neutrophil-mediated inflammatory response that contributes to the early control of colonization: (i) the pore-forming interactions of pneumolysin with epithelial cells;[42,49] (ii) the interactions of the cell-wall component lipoteichoic acid and pneumolysin with TLR-2 and TLR-4, respectively;[50] and (iii) the interactions of pneumococcal peptidoglycans with intra-epithelial nucleotide oligomerization domain (Nod)-like receptors, specifically Nod 2.[51]

In the case of adaptive immunity, IgG and secretory IgA antibodies directed against the polysaccharide capsule are generally considered to be the primary determinants of immune-mediated type-specific protection. Antibodies to pneumococcal proteins, particularly PspA, PspC (CbpA) and pneumolysin, also confer protection, which, although less efficient, is not serotype restricted.[52] Recently, innate, antibody-independent, anti-pneumococcal host defence mechanisms have been described. These are mediated by CD4+ T cells of the T-helper (Th) 1 and Th17 subsets in response to pneumococcal protein antigens.[53–55]

2. Diagnosis of Pneumococcal Infections

A large number of microbiological investigations are available, which may be used to try and identify the microbial cause of pneumonia. Among the commonly used standard investigations are sputum Gram stain and culture and pneumococcal antigen detection, and blood for culture and serological testing. Much less frequently used are invasive techniques, such as fibre-optic bronchoscopy. Yet despite the ready availability of all these investigations, the causative pathogen is only identified, at best, in approximately 50% of cases, with the greatest yield being in the more severely ill cases. Furthermore, it has been suggested that only approximately 20% of cases of pneumococcal pneumonia will be associated with bacteraemia, with the isolation of the microorganism on blood culture. A number of new techniques have been introduced, more recently with the aim of increasing the diagnostic yield for pneumococcal infection, including real-time polymerase chain reaction (RT-PCR) for rapid sputum diagnosis and rapid urine antigen testing (see section 2.1). While RT-PCR has not yet been included in most pneumonia treatment guidelines, the rapid urine test has good sensitivity and specificity but is relatively expensive, and so while it is routinely recommended in some guidelines on the management of pneumonia, other guidelines recommend it be reserved for the more severe infections, such as in cases in the ICU.

2.1 Circulating Biomarkers in Diagnosis and Assessment of Disease Severity and Outcome

Measurement of circulating pathogen-derived molecules in combination with host-derived biomarkers of infection and inflammation shows considerable promise in improving the diagnosis of invasive pneumococal disease, as well as in the assessment of disease severity and prediction of outcome. With respect to the former, the relatively recent acquisition of quantitative RT-PCR procedures for the detection of pneumococcal DNA, usually based on detection of the lyt A (autolysin) gene, in blood specimens has been reported to support the diagnosis of CAP caused by the pneumococcus, and may also be a quantitative marker of disease severity.[56–59] In the case of pneumococcal surface antigens, the BinaxNOW® S. pneumoniae immunochromatographic procedure detects the C-polysaccharide antigen in urine with good sensitivity and high specificity in adult patients with invasive disease.[58,60]

Circulating host-derived biomarkers of infection and inflammation, which are reportedly useful as diagnostic and prognostic aids, include C-reactive protein (CRP), procalcitonin, and possibly soluble triggering receptor expressed on myeloid cells-1 (sTREM-1). In the case of procalcitonin and CRP, these biomarkers, together with the circulating leukocyte count, have recently been reported to be predictive of 28-day mortality in hospitalized patients with CAP who had not received antibacterial therapy prior to presentation.[61] Measurement of sTREM-1 in bronchoalveolar lavage fluid appears to be useful in distinguishing bacterial/fungal pneumonia from viral pneumonia, atypical pneumonia, tuberculosis and non-infective inflammatory disorders.[62,63] On the other hand, measurement of circulating sTREM-1 has been reported to be of little value in the assessment of aetiology, disease severity and prediction of outcome in patients with CAP.[64]

3. Drug Resistance in Streptococcus pneumoniae

3.1 Prevalence, Evolution and Mechanisms of Antimicrobial Resistance

Numerous studies have been conducted over a considerable period of time that have documented the prevalence and mechanisms of antimicrobial resistance among isolates of S. pneumoniae. Antimicrobial resistance among pneumococci has been documented to occur worldwide and to involve the penicillins, macrolides, tetracyclines, trimethoprim, vancomycin and fluoroquinolones, as well as many other agents.[3] Virtually no antibacterial class has remained unaffected. Resistance to penicillin and other β-lactam agents has been the most discussed resistance problem, but is arguably the least important clinically, since it can usually be overcome by appropriate dosing.[65] The occurrence of drug-resistant pneumococci varies from geographical area to geographical area, and is influenced by antimicrobial prescribing habits in the different regions, and more recently has been impacted on considerably by the introduction of the pneumococcal conjugate vaccine. The latter has been associated with a significant reduction in invasive disease, as well as in colonization and infection with pneumococcal serotypes contained in the vaccine, many of which harbour resistance genes.[65] However, there has been some increase in infections with replacement serotypes, many of which now also carry antibacterial resistance.

3.1.1 Penicillin Resistance

Penicillin resistance occurs as a consequence of alterations in one or more of the cell wall penicillin-binding proteins, which catalyse bacterial cell-wall production, and this affects the affinity of the whole class of β-lactam antibacterials for these binding proteins.[65,66] This mechanism of resistance can be overcome if the concentration of the β-lactam agent at the site of infection is high enough, which allows for binding to, and inhibition of, the enzyme.[65,66] Prior antibacterial use is the prime driver of drug-resistant pneumococcal infections.[67] Penicillin-resistant pneumococci appeared in a few geographical areas, such as Australia, Spain and South Africa, in the 1970s, but subsequently spread rapidly across the world, particularly during the 1980s, 1990s and 2000s.[67,68] For example, in one study the proportion of resistant pneumococcal isolates increased nearly 30-fold during the period 1993–2004.[65] Rates of penicillin resistance exceeding 50% occur in certain areas of the world, such as Spain and Asia, but remain low (<5%) in other regions, such as Finland and Sweden.[66,67] Because such significant differences occur in different parts of the world, it is important to consider regional data when making decisions regarding antibacterial treatment.[66]

One significant change that has occurred with regard to the evaluation of penicillin resistance has been the redefining of the penicillin breakpoints for resistance in the case of non-meningeal infections. For many years penicillin susceptibility for all infections has been defined in pneumococcal strains as a minimum inhibitory concentration (MIC) <0.06 μg/mL, intermediate resistance as an MIC of 0.12–1 μg/mL and resistance as an MIC of 2 μg/mL.[65] However, in 2008, the Clinical and Laboratory Standards Institute changed the penicillin breakpoints for non-meningeal infections (such as CAP) treated with intravenous therapy as follows: susceptible ≤2 μg/mL, intermediate 4 μg/mL and resistant ≥8 μg/mL.[69] For non-meningeal infections treated with oral penicillin V (phenoxymethylpenicillin), the old breakpoints still remain valid. Using contemporary microbiological data, up to 95% of pneumococcal strains worldwide are expected to have MICs in the susceptible range for intravenous high-dose penicillin therapy.[66] Furthermore, the occurrence of highly penicillin-resistant strains with MICs ≥8 μg/mL is currently rare. Among other β-lactams, such as the cephalosporins, the MIC distribution varies for the different agents, with agents such as cefuroxime, cefotaxime and ceftriaxone being more active against the pneumococcus, although their activity has changed over the years.[65]

3.1.2 Macrolide Resistance

Macrolide resistance is most commonly mediated by one of two main mechanisms, which sometimes occur concomitantly. The first of these occurs as a result of expression of the mefA gene, resulting in the M phenotype, which is associated with an efflux pump that removes macrolides from within the cell.[65] The second occurs as a consequence of the expression of the ermB gene, associated with the MLSB (macrolide-lincosamide-streptogramin B) phenotype, which is associated with expression of an erythromycin-ribosomal dimethylase that blocks the binding of macrolides to the ribosomal target.[65] The former is associated with more moderate resistance and continued susceptibility to the lincosamides (and, therefore, clindamycin); the latter is associated with highly resistant strains, and with this mechanism there is also a block in the binding of lincosamides and streptogramin B agents.[65] The prevalence of the different macrolide resistance mechanisms varies in different parts of the world, but globally the latter is said to account for 55% overall, followed by the former in 30.6% and both in 12%.[66] Both mechanisms have been associated with failure of macrolide therapy, arguing against the use of macrolide monotherapy in areas of high prevalence of resistance.[68] The worldwide prevalence of macrolide resistance escalated at the same time as penicillin resistance, especially during the 1990s, and correlated, not surprisingly, with the use of macrolides.[67,68] Dual non-susceptibility to penicillin and macrolides has also been observed.[68] By the mid 1990s macrolide resistance exceeded 20% in many countries, but similarly to that of penicillin resistance varies by region (from <3% to >70%).[67] Clonal spread is an important vehicle for the spread of macrolide resistance.[67]

3.1.3 Fluoroquinolone Resistance

Fluoroquinolone resistance develops as a consequence of chromosomal mutations in the quinolone resistance determining region (QRDR) of the pneumococcus, involving the parC gene for topoisomerase IV and the gyrA gene of the DNA gyrase.[65] Mutations in one region may result in low-level resistance, while dual mutations confer high-level resistance.[65] Significant fluoroquinolone resistance remains uncommon, being less than 2% in most countries, but is of potential concern because of widespread use of these agents in various different settings.[65,67]

3.1.4 Multidrug Resistance

Even multidrug-resistant strains of pneumococci have emerged (resistance to three or more different classes of antibacterials), and in one study the frequency of such strains increased considerably from 9.1% in 1995 to 20% in 2005.[66]

3.2 Impact of Antimicrobial Resistance on Outcome of Pneumococcal Infections

While many studies have investigated the prevalence and mechanisms of pneumococcal resistance, much less attention has been focused on the true impact of antimicrobial resistance on the outcome of pneumococcal infections treated with standard antimicrobial agents. In many studies, current levels of antibacterial resistance have been shown to have very limited impact on the clinical outcomes of patients with pneumococcal CAP, particularly with regard to penicillin and other β-lactam agents, yet antimicrobial prescribing habits have changed because of concerns about resistance.[70–72] There is also mounting evidence that supports relatively simple strategies to overcome the impact of resistance, such as using high doses of antimicrobial agents, using more active agents within a specific class of antibacterials or switching to another class of antibacterial, particularly in cases at increased risk of infection with highly resistant pneumococci.[73]

3.2.1 β-Lactam Resistance

A large, multicentre, prospective, international, observational study investigated 844 hospitalized patients with S. pneumoniae bacteraemia.[74] The investigators were specifically chosen since they worked in institutions in cities or countries that had previously been reported to have a high prevalence of drug non-susceptible pneumococci. Overall, 15% of isolates were of intermediate susceptibility to penicillin (MIC 0.12–1 μg/mL) and 9.6% were highly resistant (MIC >2 μg/mL). The impact of concordant antibacterial therapy (receipt of one antibacterial with high in vitro activity against the pneumococcal isolate) versus discordant therapy (antibacterial inactive in vitro) on 14-day mortality was assessed. Discordant therapy with penicillins, cefotaxime and ceftriaxone (but not cefuroxime at 750 mg three times daily — see section 3.4) did not result in a higher mortality. Neither was there a difference in time to defervescence or frequency of suppurative complications. An additional study indicated that only discordant therapy with β-lactam agents that had poor anti-pneumococcal activity impacted on outcome, but not penicillins or broad-spectrum β-lactams.[75] Furthermore, although there are some studies indicating a possible impact of β-lactam resistance on outcome, a critical review of the literature among patients with pneumococcal pneumonia, both with and without bacteraemia, revealed only one single case of a documented microbiological failure of parenteral penicillin-class antibacterials in a patient with an empyema, into which antibacterials are known to penetrate poorly.[76] Some have therefore suggested that even penicillin appears to be adequate and effective, when administered in adequate dose and frequency, for the treatment of pneumococcal pneumonia.[77]

3.2.2 Macrolide Resistance

The same is not quite true for the macrolides and fluoroquinolones.[73,78,79] With regard to the macrolides, Lonks and colleagues[78] conducted a matched case-control study of patients with bacteraemic pneumococcal infections to determine whether breakthrough bacteraemia occurring during macrolide treatment was related to macrolide susceptibility of the pneumococcal isolates. Cases were patients with pneumococcal bacteraemia and isolates that were either resistant or intermediately resistant to erythromycin, whereas controls were age-, sex-, location- and year-matched cases in whom the isolates were susceptible to erythromycin. Excluding meningitis cases, 18 of the cases (24%) and none of the 136 matched controls were taking a macrolide when blood was taken for the culture (p = 0.00000012). A similar result was seen even in cases infected with isolates carrying the low-level resistant M phenotype, thus suggesting that both efflux and methylation mechanisms of resistance may be clinically relevant and associated with breakthrough bacteraemias in patients being treated with macrolides.[78–80] A number of additional cases of macrolide treatment failures have been reported, although these are few in number compared with the overall number of pneumococcal cases, and specifically cases due to macrolide-resistant isolates seen every year.[73]

3.2.3 Fluoroquinolone Resistance

Similarly, fluoroquinolone treatment failures have also been documented in patients infected with fluoroquinolone-resistant S. pneumoniae.[73] All of these cases were treated with either ciprofloxacin, known to be poorly active against the pneumococcus or levofloxacin given as 500 mg daily, a dose that is not considered to be optimum based on a current understanding of optimal pharmacokinetic (PK)/pharmacodynamic (PD) parameters (see section 3.4).

3.3 Antibacterial Therapy of Drug-Resistant Infections

In a number of review articles, the evidence in the literature for the impact of antimicrobial resistance on the outcome of pneumococcal infections, treated with various different antibacterial classes, has been evaluated and firm recommendations made with regard to specific therapy.[81–84] These are described more fully in section 4. In the first instance it is important to note that it is not only the specific antibacterial class itself that is important in the outcome of an infection, but also the dose that is given, the route it is administered, the timeliness of antibacterial administration from the time of presentation of the patient with the infection, and the PK/PD properties of the agent administered (see section 3.4).

In general, certainly with regard to the use of parenteral β-lactam agents for the treatment of pneumococcal pneumonia, current prevalence and levels of penicillin resistance are such that either penicillin itself or the aminopenicillins could be used in standard doses. In the case of infections with strains of intermediate resistance, higher drug dosages are recommended. In the case of high-level resistance, alternative agents, such as the third-generation cephalosporins (e.g. ceftriaxone, cefotaxime) or the respiratory fluoroquinolones (e.g. levofloxacin, moxifloxacin, gemifloxacin), should be used.

With regard to the macrolides (including the azalide agent azithromycin), these agents are not recommended for use as monotherapy in areas where there is a high prevalence of macrolide-resistant pneumococcal infections. However, in the case of a young, previously healthy individual who has not had a recent course of antibacterials and is presenting with a mild pneumonia that is to be treated at home, macrolide monotherapy may be suitable, since in this situation macrolide resistance is much less likely to occur. Furthermore, macrolides are still recommended as appropriate therapy for so-called ‘atypical’ infections and also as part of combination therapy. The latter (i.e. β-lactam/macrolide combination) is considered a suitable therapy for the hospitalized patient with more severe pneumococcal CAP, including the subset of cases with pneumococcal bacteraemia. An alternative choice for the latter is fluoroquinolone monotherapy, although it is often recommended that these agents are reserved for specific cases in order to prevent rapid development of antibacterial resistance.

With regard to the fluoroquinolones, while the presence of two significant mutations in the QRDR is associated with fluoroquinolone resistance, such that these agents are not suitable for therapy, this occurrence is still very uncommon worldwide. However, what is not fully appreciated is that isolates with single-step mutations do occur, may well test as susceptible in the laboratory, and resistance may subsequently occur during fluoroquinolone therapy due to the spontaneous occurrence of a second mutation. Furthermore, the prevalence of these single-step mutations in many areas of the world is uncertain. When used, fluoroquinolones need to be given in appropriate doses that may limit the emergence of these mutations (see section 3.4).

3.4 Pharmacokinetic/Pharmacodynamic Parameters and Antibacterial Choice andDosing

The use of drug PK/PD principles is the new pharmacological science that enables us to understand the relationship between drug dosing and its likely efficacy, and is particularly useful in the era of emerging antibacterial resistance. This has been discussed in a number of review articles and is described more fully below.[85–87] With the use of the PK parameters of drug serum concentration over time and area under the concentration-time curve, and integrating these with the MIC of the microorganism, one could predict the likelihood of clinical success and pathogen eradication. Use of PK/PD parameters is also helpful for preventing selection and spread of resistant microorganisms and has led to the development of the concept of the mutant selection concentration, which is the lowest concentration of the antimicrobial that prevents selection of resistant bacteria from high inocula of organisms.[88]

β-Lactam antibacterials and the macrolides (but not the azalides) are time-dependent antimicrobials and the major PK/PD parameter correlating with the outcome is the so-called time above MIC (T > MIC); the serum concentration of the antibacterial needs to be above the MIC of the microorganism for 40–50% of the dosing interval for likely success. Using standard dosing regimens of the various drugs and comparing these to the MIC(s) of individual pathogens or a collection of strains, one can determine if a T > MIC of 40–50% of the dosing interval is achieved (equivalent to the breakpoint of the pathogen being below the resistance breakpoint) and, therefore, whether use of the drug is likely to be associated with clinical success. It is for this reason that there is continuous ongoing success with the use of the penicillins and aminopenicillins in the management of pneumococcal infections in most areas of the world, since given the current levels of pneumococcal penicillin resistance worldwide, together with appropriate increased dosing, a T > MIC of 40–50% or greater is readily achieved. In the case of cefuroxime, the T > MIC with standard dosing is borderline, particularly in the presence of slightly elevated MICs, but sufficient with higher dosing (e.g. parenteral cefuroxime 750 mg three times daily has been associated with treatment failures but not 1500 mg three times daily[74]). In the case of the macrolides, a T > MIC of 40–50% is achieved with susceptible isolates, but not with macrolide-resistant isolates, in which macrolide monotherapy, in any dose, is therefore not recommended. These agents may still be used as part of combination therapy (see section 4.2).

In the case of the fluoroquinolones, which kill pathogens by a concentration-dependent mechanism, the major PK/PD parameter predictive of likely outcome is the so-called AUIC (area under the inhibitory curve = the area under the serum drug concentration curve to MIC ratio). While there are differences in the values predictive for likely failure or success of fluoroquinolone therapy in immunocompetent versus immunosuppressed individuals, in milder or more severe infections and for Gram-negative versus Gram-positive pathogens, a number of studies suggest (as do many investigators) that the appropriate AUIC value to aim for in both Gram-positive and Gram-negative pathogens should be similar (>100 or even >125).[89] This is not achieved with either oral or parenteral dosages of levofloxacin of 500 mg/day, but is achieved with 750 mg/day, which is the currently recommended dose for this fluoroquinolone and is also achieved with standard doses of moxifloxacin[90] and gemifloxacin.

4. Antibacterial Therapy of Pneumococcal Pneumonia

4.1 Early Initiation of Antibacterial Therapy

A number of studies have suggested that there is a significant and causal relationship between timing of initial antibacterial therapy and improved outcome (length of hospital stay and mortality) in patients with CAP.[91] This association appeared to be particularly strong among older patients who had not yet received antibacterial prior to arriving at the hospital. However, two recent studies among patients with bacteraemic pneumococcal CAP have indicated that administration of adequate antimicrobial therapy within 4 hours of arrival at hospital was a critical determinant of survival in these patients.[92,93] In the former study, 363 patients were studied. The median time to first administration of antibacterials was 2.8 hours. Overall, 66% of patients received at least one active antibacterial within 4 hours, 82% within 8 hours and 94% within 24 hours. Receipt of at least one active antibacterial was associated with a reduced mortality (odds ratio [OR] 0.47 [95% CI 0.2, 1.0]; p = 0.04) and shortened length of stay (OR 0.77 [95% CI 0.60, 1.0]; p = 0.03). In the latter study, a time period of >4 hours to the first administration of adequate antibacterials was independently associated with in-hospital mortality.[93] As a result of the many studies, time to first antibacterial administration in patients presenting to hospital has been an audited performance measure for CAP for many years.[94–97] However, there have been some concerns about the recommendation of antibacterial administration within 4 hours of presentation of patients with suspected CAP. First, this would necessitate the treatment of at least some patients, such as those presenting in an atypical manner, before a firm diagnosis of CAP is made.[95,96] Second, in some studies that have shown a benefit of early antibacterial administration (within 4 hours) on outcome, the factors associated with antibacterial delay were conditions such as altered mental state, absence of fever, absence of hypoxia and increasing age, many of which, in themselves, may impact negatively on mortality.[95,96] The Infectious Diseases Society of America (IDSA)/American Thoracic Society (ATS) pneumonia guideline now recommends that the first antibacterials be given in the emergency department, rather than assigning a specific timepoint to this process.[98]

4.2 Combination Antibacterial Therapy

A myriad of studies in patients with CAP, both of all causes (including cases of pneumococcal infections) as well as those due to S. pneumoniae alone (including the subset of patients with pneumococcal bacteraemia), have indicated that combination antibacterial therapy, most commonly the addition of a macrolide to standard β-lactam antibacterial therapy, is associated with improved outcomes[99–103] (table I). Furthermore, the benefits of adding a macrolide to therapy in patients with CAP extended to cases with severe sepsis, as well as to intubated patients.[108,109] These findings need to be counterbalanced by additional studies not showing such benefits, or showing benefits in only selected subgroups of patients,[104–107] as well as by prospective, randomized investigations suggesting that fluoroquinolone monotherapy may be at least as effective as combination therapy.[110,111] This situation is further confounded by other studies indicating that the use of a β-lactam/fluoroquinolone combination in patients with severe pneumonia of all causes may be associated with increased short-term mortality compared with that of other, guideline-compliant therapies.[112] As a consequence of these various studies, and despite the apparent contradictions, most guidelines, such as the IDSA/ATS guideline,[98] recommend the use of a β-lactam/macrolide combination or fluoroquinolone monotherapy for the treatment of sicker, hospitalized patients with CAP, including the subset of cases with pneumococcal bacteraemia.[98,113] It has been suggested that the major discriminatory factor that may influence the choice between fluoroquinolone monotherapy or β-lactam/macrolide combination therapy is the history of recent prior antibacterial therapy in the patient.[98] Another factor that may be of influence is a history of antibacterial allergy in the patient. One additional question that has been raised is how long the benefit from combination therapy lasts and, therefore, how long combination antibacterial therapy should be continued. In the study by Baddour and colleagues,[102] the potential benefit of combination antibacterial therapy was evaluated for both day 1 and day 3 and was found to be present for both. It is usually recommended that combination therapy be continued for at least 3 days. Another consideration is what to step down to or switch to (see section 4.3) when intravenous combination therapy with a β-lactam and a macrolide has been used initially. It has been suggested that it may be to either class of drug, including a macrolide alone, provided the patients are not infected with drug-resistant S. pneumoniae or Gram-negative enteric pathogens.[98]

Table I
figure Tab1

Studies documenting the impact of combined antibacterial therapy on the outcome of pneumococcal pneumonia, predominantly in cases with pneumococcal bacteraemia

Full size image

The exact reason(s) for and/or mechanism(s) of benefit of the addition of macrolides is uncertain, but may be multifactorial (see later in this section, section 5.1 and tables II and III). In the study by Gamacho-Montero and colleagues[93] described previously, combination therapy was protective against delayed adequate therapy (adjusted hazard ratio [HR] 0.53 [95% CI 0.29, 0.95]; p = 0.033), the latter being potentially associated with a poorer outcome. Another suggestion is that the addition of macrolides would cover for so-called ‘atypical pathogens’. Interestingly, Metersky and colleagues[115] addressed the question of whether adding agents active against ‘atypical pathogens’ (namely macrolides, fluoroquinolones or tetracycline) was associated with better outcome in patients with bacteraemic pneumonia. Their study indicated that while the initial use of an antibacterial active against atypical pathogens was independently associated with decreased risk of 30-day mortality and hospital admission within 30 days of discharge, this benefit was only associated with the use of macrolides and not fluoroquinolones or tetracyclines. There is additional evidence to suggest that the beneficial effects of macrolides may go beyond their primary antimicrobial activity. For example, in the study by Restrepo and colleagues[108] of patients with severe pneumonia and sepsis, benefit was seen with the addition of macrolides even in the presence of macrolide-resistant microorganisms.

Table II
figure Tab2

Possible mechanism(s) of benefit associated with the addition of a macrolide to standard β-lactam therapy in the management of patients with pneumococcal community-acquired pneumonia (reproduced from Feldman and Anderson,[2] with permission)

Full size image
Table III
figure Tab3

Pathogen- and host-directed anti-inflammatory activities of macrolides

Full size image

A further mechanism may relate to anti-inflammatory effects that combination antibacterial therapy has on cytokine release. In severe pneumococcal pneumonia, acute-phase proteins and various cytokine levels are raised and the longer the time from onset of pneumonia symptoms to hospital presentation, the higher these values are.[116]

Furthermore, levels of TNF correlate with levels of IL-1β, IL-6 and IL-8 and are associated with the presence of bacteraemia, initial blood pressure <90 mmHg and lower oxygen concentration on admission, all potential indicators of severity. In subsequent studies, high IL-6 levels were associated with the worst outcomes in patients with pneumococcal pneumonia and initial combination antibacterial therapy produced a faster decrease in IL-6 levels than monotherapy.[117] These and a range of other, non-antimicrobial, anti-inflammatory, immunomodulatory effects of macrolides are believed by many to underlie the benefits achieved with combination therapy.[118,119]

Guideline-compliant therapy, such as is indicated in the IDSA/ATS guideline, with the use of a β-lactam/macrolide combination or fluoroquinolone monotherapy has been studied and been shown to be associated with lower mortality, decreased complication risk, decreased time to clinical stability and associated duration of parenteral therapy, and decreased length of hospital stay and therefore overall resource utilization in adult patients, including the elderly, with CAP.[120,121] Furthermore, the use of a macrolide/β-lactam combination or fluoroquinolone monotherapy in patients hospitalized with CAP has been included as one of the quality measures in the treatment of patients with CAP.[97] The ultimate choice of the antibacterial regimen for the individual patient (i.e. β-lactam/macrolide combination or fluoroquinolone monotherapy) would depend on a number of host factors, including an appreciation of what antibacterials the patient has had in the recent past (the preceding 90 days), the presence of allergy to a particular class of antibacterials and/or other factors that may preclude the use of certain agents. The reason that preceding antibacterial use should be taken into consideration is that it increases the likelihood of the current infection being due to microorganisms that are resistant to that previously used class of antibacterials.[98]

4.3 Switch and De-Escalation Therapy

For those patients with CAP who are admitted to hospital, most guidelines recommend that patients should be switched from intravenous to oral antibacterial therapy (‘switch therapy’) as soon as they are haemodynamically stable, are clinically improving, are able to ingest oral medications and have no gastrointestinal dysfunction.[98] Ramirez and colleagues[122,123] established criteria for early switch therapy, which are commonly used. In the IDSA/ATS CAP guideline,[98] the criteria indicated for clinical stability include a temperature <37.8°C, heart rate <100 beats/min, respiratory rate <24 breaths/min, systolic blood pressure >90 mmHg, arterial oxygen saturation >90% or partial pressure of arterial O2 (PaO2) >60 mmHg on room air in a patient who is able to maintain oral intake and has a normal mental status. Subsequent studies have suggested that even more liberal criteria are adequate for switch to oral therapy.

Ramirez and colleagues[124–126] studied early switch therapy in a number of clinical situations, including both CAP of all cause as well as bacteraemic community-acquired S. pneumoniae pneumonia. In the earlier study by Ramirez and Bordon,[125] patients with pneumococcal bacteraemia were less likely to reach clinical stability and become candidates for switch therapy than general populations of CAP patients, and there was also a delay in time to reach clinical stability. However, in the absence of meningitis or endocarditis, bacteraemic patients reaching clinical stability could safely be stepped down to oral therapy.[125] This study was superseded by a more recent study from this research group, which was a secondary analysis of the Community-Acquired Pneumonia Organization (CAPO) database of hospitalized patients with CAP and pneumococcal bacteraemia (124 cases).[126] The initial association between pneumococcal bacteraemia and poorer outcomes became insignificant when adjusting for other co-variates. Thus, the multivariate regression analysis revealed no association between bacteraemic CAP and time to clinical stability (HR 0.87 [95% CI 0.7, 1.1]; p = 0.25), length of hospital stay (HR 1.14 [95% CI 0.91, 1.43]; p = 0.25), all-cause mortality (OR 0.68 [95% CI 0.36, 1.3]; p = 0.25) or CAP-related mortality (OR 0.86 [95% CI 0.35, 2.06]; p = 0.73). Clearly, the factors related to severity of illness were confounders for the association between pneumococcal bacteraemia and poor outcome, explaining the earlier findings. The authors concluded that pneumococcal bacteraemia itself was not a contraindication to de-escalating therapy in clinically stable patients.

However, despite these findings, a more recent study has documented that there is evidence in the literature of considerable variability in the practice of early switch therapy for patients with CAP.[127] This needs to be addressed since the advantages of early switch therapy are that patients are converted from intravenous to oral therapy earlier and usually discharged from hospital sooner. Since duration of parenteral antibacterial therapy is often the primary factor affecting length of hospital stay, and length of hospital stay is the major determinant of costs of therapy, early switch therapy and early discharge may be associated with significant cost saving in the management of patients with CAP.

5. Adjunctive Therapies for Pneumococcal Community-Acquired Pneumonia

As mentioned earlier, β-lactam antimicrobial agents are the cornerstone of therapy for pneumococcal pneumonia. Nonetheless, considerable effort continues to be directed at the identification of adjunctive therapies that attenuate adverse inflammatory responses or, alternatively, augment host defences. Foremost among the former are macrolide antibacterials, largely because of their secondary anti-inflammatory properties, while corticosteroids and possibly cyclic adenosine monophosphate (cAMP)-elevating agents show promise. The latter group includes passive immunotherapeutic agents such as hyperimmune serum, intravenous gammaglobulin and monoclonal antibodies. The therapeutic potential of inhibitors of intravascular coagulation, exogenous surfactant and statins (HMG-CoA reductase inhibitors) has recently been reviewed in detail elsewhere.[128]

5.1 Macrolides

While macrolides are more usually considered simply as antimicrobial agents that have a role in the antibacterial therapy of pneumococcal CAP, the mechanism(s) underlying their benefit may not relate entirely to their antimicrobial activity but rather to their additional activities (table III). As such, although not yet conclusively proven, inclusion of a macrolide may actually represent the most compelling adjunctive strategy in the treatment of severe pneumococcal pneumonia.[115,129] Macrolides possess a combination of properties, both antimicrobial and non-antimicrobial, that are likely to underpin their apparent usefulness as adjuncts to β-lactams in CAP. Benefit related to antimicrobial activity appears to result from the bacterostatic and protein synthesis inhibitory effects of these agents.[115] The benefits of non-antimicrobial activity are largely attributable to the immunomodulatory and anti-inflammatory activities of macrolides[114,129] (table III).

5.1.1 Indirect Anti-Inflammatory Activity of Macrolides

Bactericidal antibacterials, including β-lactams and fluoroquinolones, exacerbate pathogen-directed inflammatory responses as a consequence of the release of proinflammatory intracellular toxins and cell-wall components from disintegrating bacteria, which is likely to be most evident in the clinical setting of high bacterial loads. In the case of the pneumococcus, release of lipoteichoic acid and peptidoglycan from the cell wall may exacerbate the inflammatory response via interactions with TLR-2 and Nod 2, respectively, which is intensified by the pore-forming, TLR-4-binding and complement-activating effects of pneumolysin. On the other hand, the actions of inhibitors of bacterial protein synthesis are more subtle and controlled. These antibacterials, especially macrolides and macrolide-like agents, subdue and weaken their target pathogens by attenuating the production of proinflammatory protein toxins and other virulence factors such as adhesins, quorum sensors and biofilm. Importantly, these activities of macrolides are not only evident with macrolide-susceptible strains of the pneumococcus, but also with macrolide-resistant strains, as well as organisms with innate resistance such as Escherichia coli and Pseudomonas aeruginosa.[130–134] In the case of the pneumococcus, we have found macrolides to be extremely effective inhibitors of the production of pneumolysin, even in the setting of macrolide resistance.[131,132] The importance of this is that pneumolysin is considered by many to be the most important virulence factor of the pneumococcus and plays a major role in the pathogenesis of severe pneumococcal pneumonia (see section 1.2.2).

The distinction between β-lactams and macrolide/macrolide-like agents with respect to proinflammatory activity has been demonstrated in several models of experimental infection, including a recent study using a murine model of secondary, influenza-associated pneumococcal pneumonia. In this study, the lowest survival rate in antibacterial-treated animals was observed in those treated with ampicillin only, with the highest rates being observed in those treated with azithromycin or clindamycin only, or in combination with ampicillin.[135] Improved survival in the groups treated with azithromycin or clindamycin was associated with an attenuated inflammatory response, demonstrating that macrolides counteract the pathogen-directed proinflammatory activity of β-lactams.

5.1.2 Direct Anti-Inflammatory Activities ofMacrolides

Macrolides have extremely high levels of tissue penetration, and are highly concentrated by epithelial cells and cells of the innate immune system. These agents appear to be particularly effective in controlling neutrophil-mediated inflammation, which may explain their efficacy in the therapy of acute and chronic respiratory disorders such as COPD, panbronchiolitis, obliterative bronchiolitis and cystic fibrosis in which the neutrophil appears to be the primary offender (reviewed by Feldman and Anderson[114]). Several mechanisms of anti-inflammatory activity, which are possibly interactive, have been attributed to macrolides. These include membrane-stabilizing activity,[136] as well as inhibition of synthesis of the potent neutrophil chemoattractant IL-8 by a variety of structural and inflammatory cells, including bronchial epithelial cells and monocytes (reviewed in Feldman and Anderson[137]). This latter activity results from macrolide-mediated interference with intracellular signalling mechanisms that converge on transcriptional activation of IL-8 gene expression.[138–140]

In summary, the unusual combination of excellent cell and tissue penetration, antimicrobial and anti-inflammatory activities appear to account for the apparent efficacy of macrolides as adjuncts to β-lactams in the treatment of severe pneumococcal pneumonia, as opposed to activity against atypical pathogens.[115]

5.2 Corticosteroids

Adjunctive corticosteroids have become routine treatment in the clinical management of adults with bacterial meningitis, significantly reducing hearing loss and neurological sequelae, as well as mortality in pneumococcal meningitis.[141,142] To date, however, there are no published studies that have specifically addressed the adjunctive potential of corticosteroids in severe pneumococcal pneumonia. Several relatively small studies, most recently those reported by Confalonieri et al.[143] and Garcia-Vidal et al.[144] have reported a benefit of early administration of systemic corticosteroids to hospitalized patients with severe CAP. Clinical benefit manifested as significant improvements in the PaO2/FiO2 (fraction of inspired O2) ratio and chest radiograph, as well as reductions in the multiple organ dysfunction score and mortality. However, in a recent and much larger randomized, double-blinded clinical trial, Snijders et al.[145] did not detect beneficial effects of early administration of corticosteroids (systemic or oral) on outcome of patients with CAP. Although the authors conceded that possible benefit of adjunctive corticosteroids in more severely ill patients could not be excluded, it is noteworthy that Sprung et al.[146] also failed to detect a benefit of intravenous corticosteroids in patients with septic shock, irrespective of the response to corticotrophin. However, hydrocortisone therapy did hasten reversal of shock, but this was negated by the higher frequency of superinfection, including new sepsis and shock, in the corticosteroid-treated group.[146]

On the basis of recent evidence, the role of corticosteroids in the adjunctive therapy of severe pneumococcal pneumonia remains uncertain. However, it is noteworthy that corticosteroids, unlike macrolides, are relatively ineffective in controlling the harmful proinflammatory activities of neutrophils,[147] suggesting that these agents may be most effective when they are used in combination. In this respect, it may be meaningful that “the use of macrolides was discouraged because of their immunomodulating effect” in the study reported by Snijders et al.,[145] while Confalonieri et al.[143] “followed the 1993 American Thoracic Society Guidelines for the initial management of adults with community-acquired pneumonia”, which advocates (i) “a second- or third-generation cephalosporin or beta-lactam/beta-lactamase inhibitor +/- macrolide for hospitalized patients with CAP”; and (ii) a “macrolide + third-generation cephalosporin with anti-Pseudomonas activity or other anti-pseudomonal agents such as imipenem/cilastatin, ciprofloxacin for severe hospitalized patients with community-acquired pneumonia”.[148] Antibacterial usage was not specified in the reports authored by Garcia-Vidal et al.[144] and Sprung et al.[146]

5.3 Cyclic Adenosine Monophosphate-Elevating Agents

cAMP possesses broad-ranging, anti-inflammatory activities affecting various types of immune and inflammatory cells and their proinflammatory mediators, and has been described recently as the “master regulator of innate immune cell function”.[149] The molecular/biochemical basis of the anti-inflammatory activity of cAMP largely involves activation of cAMP-dependent protein kinase A. This kinase, in turn, mediates the removal/exclusion of Ca2+ from the cytosol of activated immune and inflammatory cells by several interactive mechanisms,[150] and also antagonizes the interaction of NF-κB with the transcriptional cofactor cAMP response element binding-protein (CREB), a critical event in activation of histone deacetylase and gene expression.[151] cAMP-mediated clearance of cytosolic Ca2+ effectively downregulates the Ca2+-dependent proinflammatory activities of neutrophils, including the generation of reactive oxygen species and LTB4, expression of the β2-integrin CR3 and release of granule proteases.[152–154] Antagonism of NF-κB, on the other hand, results in decreased synthesis of proinflammatory cytokines, especially IL-8 and TNF, by other cell types such as monocytes/macrophages and epithelial cells.[151]

Although largely untested in severe pneumococcal pneumonia in either a clinical or experimental setting, sepsis has been identified as being a potential area for the therapeutic application of cAMP-elevating pharmacological agents, and several experimental studies appear to bear this out. Importantly, human leukocytes possess G-protein-coupled adenosine A2A, β2-adrenergic and E series of prostaglandins (EP) receptors, all of which are linked to adenylyl cyclase; they also possess cyclic nucleotide phosphodiesterase (PDE) enzymes (reviewed by Tintinger et al.[150]).

In patients with severe sepsis/septic shock, intracellular cAMP levels are significantly decreased in blood mononuclear leukocytes, which is associated with impairment of both β-adrenergic receptor-dependent and -independent activation of adenylyl cyclase, and an extended post-receptor defect of β-adrenergic signal transduction.[155] Decreased intracellular cAMP is likely to result in hyper-reactivity of immune and inflammatory responses. In addition, defective β-adrenergic signalling in the cardiovascular system in humans appears to underpin the myocardial hypo-responsiveness to catecholamines/myocardial depression that occurs in sepsis.[156,157]

Notwithstanding the use of inotropes, strategies to augment intracellular cAMP in the acute clinical setting are, realistically, limited to non-methylxanthine, nonspecific inhibitors of PDEs. This is because leukocytes and structural cells vary with respect to their expression of the various PDE subtypes, clearly restricting the use of selective inhibitors, while methylxanthines are potentially toxic. Three agents, all of which are non-methylxanthine, nonspecific PDE inhibitors, merit serious consideration as potential adjuncts in the therapy of severe pneumococcal disease and sepsis/septic shock. These are pentoxifylline, which has already shown promise in neonatal sepsis,[158–160] as well as ibudilast[161] and montelukast.[153] Neither ibudilast nor montelukast has been evaluated in patients with severe CAP/sepsis. However, both agents combine cysteinyl LT receptor antagonistic activity with nonspecific PDE inhibitory activity,[153,161] making them particularly attractive contenders for evaluation in this setting.

5.4 Antibody Administration

In his recent review, Wunderink[128] describes the small reduction in mortality (~10%) that was associated with the passive administration of hyperimmune serum to patients with pneumococcal pneumonia in the pre-antibacterial era, while the benefit, if any, of administration of intravenous gammaglobulin to patients with CAP remains unproven. Such a study was conducted several years ago by one of us, among patients with suspected pneumococcal CAP admitted to an ICU in Johannesburg, South Africa (Feldman C, unpublished data). This was a prospective, randomized, double-blind, placebo-controlled study using a hyperimmune pneumococcal gammaglobulin preparation containing antibodies to 14 pneumococcal serovars/groups administered in a dose of 400 mg/kg or matching placebo, which was administered to all study patients within 24 hours of admission to hospital/ICU. In addition, all patients received identical standard treatment for CAP, including antibacterials, as per the ICU protocol. An attempt was made to include only cases with pneumococcal CAP, using strict microbiological criteria, and additional laboratory testing included sputum Gram stain and culture, blood cultures, and sputum and blood countercurrent pneumococcal immunoelectrophoresis. Twelve cases were initially enrolled in the study, of which nine were subsequently confirmed to have pneumococcal infection. Of these nine pneumococcal cases, four of the six serum-treated patients died, whereas all three placebo-treated patients survived (p = 0.12). A further three cases of suspected pneumococcal pneumonia, not subsequently confirmed as having pneumococcal infection, had been enrolled, one of whom had received serum and subsequently died. The other two survived. Thus, a total of five of 12 patients died, all of whom had received serum, and seven survived, two of whom had received serum (p = 0.027). On the basis of these differences in mortality, the study was stopped as was a requirement of the regulatory authorities in South Africa.

An alternative, experimental approach described by Garcia-Suárez et al.[162] was based on the intravenous administration of monoclonal antibodies against pneumolysin to mice experimentally infected with the pneumococcus. Although immunotherapy was associated with a decrease in bacterial lung colonization and lower frequencies of tissue injury and bacteraemia, it may be expensive and impractical in the clinical setting.

6. Conclusions

Bacteraemic pneumococcal pneumonia continues to have a major medical impact throughout the world. Much recent research has focused on optimal strategies for the management of this condition. Antimicrobial therapy is potentially compromised by emerging resistance of this microorganism to commonly used antibacterials. However, a greater understanding of PK/PD parameters, together with knowledge derived from various clinical studies, have allowed us to choose suitable agents or combinations of agents, in appropriate dosages, that are most commonly associated with a better outcome. A number of adjunctive therapies have also been studied in an attempt to further reduce the high mortality of pneumococcal infections, of which the macrolides themselves, and possibly the corticosteroids, appear to be the most promising. Much further research is still needed and is currently ongoing.