Had we lived in a culture with a greater awareness of history, the sense of security engendered by the initial success with penicillin might have been tempered by earlier events. Optochin, a drug related to quinine, was used briefly in the second decade of this century to treat pneumococcal infection until its ocular toxicity forced its abandonment. Reports of pneumococci showing increased resistance to optochin recovered from mice treated experimentally with this drug appeared in 1912 [5], and similar strains were isolated from the blood of two patients receiving optochin at the Hospital of the Rockefeller Institute several years thereafter [6, 7]. Twenty years later, reports of pneumococci resistant to sulfonamides appeared shortly after the introduction of these drugs [8].

Clear demonstration of the ability of pneumococci to give rise in vivo to mutants showing increased resistance to penicillin was reported by Schmidt and Sesler in 1943, the year before the publication of Tillett and colleagues, an observation confirmed in vitro 2 years later by Eriksen [10]. Two decades were to elapse before isolation of similar strains from humans was described [11, 12]. The situation is not dissimilar from that of sulfonamide resistance in meningococci, which was observed in the laboratory in 1949 by Branham [13] and not noted to be a problem in humans until the 1960s [14]. It is probably a truism that, if mutants of an organism resistant to an antibacterial agent can be isolated in the laboratory, sooner or later similar resistant mutants will be isolated from humans, although protracted periods of time may elapse between the two observations.

Pneumococcal resistance to penicillin results from alterations in the genetic structure of the organism, giving rise to changes in one or more of its penicillin-binding proteins, thereby reducing affinity for the drug [15]. Several possible mechanisms have been proposed for such alterations, including the selection of spontaneous mutants and the exchange of the DNA of a sensitive strain for the DNA from a resistant strain of pneumococcus or from a streptococcal component of the normal respiratory flora that is relatively insensitive to penicillin [16]. The studies of Hotchkiss [17] and of Shockley and Hotchkiss [18] showed clearly the genetic basis of penicillin resistance in pneumococci, demonstrating the transfer of multiple increments of resistance to an initially sensitive strain with the DNA of a highly resistant strain. In addition, there is now considerable circumstantial evidence, based on studies of multiple characteristics of the organism, to indicate that spread of resistant mutants to geographically distant areas may be accomplished by today's rapid transport and movement of people throughout the world [19].

Because of the incremental nature of pneumococcal resistance to penicillin, arbitrary limits have been established to define sensitive and resistant strains. Those strains inhibited under standard cultural conditions by 0.1 µg/mL or less are classified as sensitive, those inhibited by concentrations between 0.1 µg/mL and 2 µg/mL are said to manifest intermediate resistance, and those requiring concentrations in excess of 2 µg/mL are said to be fully resistant to the drug [20]. Strains growing in concentrations of penicillin in excess of 8 µg/mL have been isolated from humans and capsulated strains requiring concentrations as high as 12 to 20 µg/mL for their inhibition have been selected in the laboratory [21]. An additional feature of some pneumococcal strains resistant to penicillin is their manifestation of resistance to one or more additional antipneumococcal agents, including sulfonamides, cotrimoxazole, macrolides, tetracyclines, chloramphenicol, and aminoglycosides. Strains resistant to three or more antimicrobial agents having different mechanisms of bacterial inhibition are defined as multidrug-resistant [20]. Multidrug-resistant pneumococci were first identified in South Africa in 1977 [22]. The only drug of utility in treating pneumococcal infection to which drug resistance of the causal organism has not yet been recognized, with one possible exception [23], is vancomycin.

Since the initial identification of penicillin-resistant and of multidrug-resistant pneumococci, strains of both kinds have been isolated from asymptomatic carriers and from infected individuals from all five continents [24]; in Spain in 1989, strains highly resistant to penicillin constituted as much as 20% of the strains recovered from blood [25]. In the United States, pneumococci manifesting intermediate and high-level resistance to penicillin constitute 5% to 10% of isolates examined in hospital laboratories [26], although local areas with higher proportions of resistant strains exist [27]. In light of these findings, empiric treatment of a patient with a serious pneumococcal infection can no longer be regarded as an acceptable practice, and quantification of the susceptibilities of the infecting organism to drugs of potential utility has become essential. Although any of the 84 capsular types of pneumococci may develop resistance, the preponderance of strains manifesting this property can be found among serotypes or groups 6, 9, 14, 18, 19, and 23, strains of which are isolated most commonly from children, who are treated frequently with antimicrobial drugs [28]. Capsular serotyping by the Quellung reaction provides a rapid means of identifying pneumococci and is especially helpful when bacteria can be visualized in cerebrospinal fluid. For this purpose, a reagent called "Omniserum" [29], prepared by the Danish Statens Seruminstitut, which reacts with all known pneumococcal capsular types, is most useful and is available in the United States.

In light of the findings just described, how should the contemporary treatment of pneumococcal infection be determined? As noted earlier, pulmonary infections caused by "wild type" pneumococci manifesting their usual sensitivity to penicillin G can be treated conventionally with doses of 300 000 to 600 000 units of the drug given intramuscularly four times daily. For the initial treatment of those who are gravely ill with pneumonia or other pneumococcal infections, with the exception of those of the central nervous system, penicillin G may be given but in significantly larger doses than those conventionally used. To maintain a serum level of penicillin in excess of 5 µg/mL, Tucker and Eagle [30, 31] have ascertained that 1 200 000 units or 10 mg/kg of body weight must be given parenterally every 2 hours (approximately 14 000 000 units a day). Such a regimen should provide levels adequate to inhibit highly resistant strains until laboratory determination of the sensitivities of the infecting organism has been completed. A similar regimen has been proposed by Pallares and colleagues [32] for the management of infections caused by pneumococcal strains resistant to penicillin in concentrations of 2 µg/mL or less. Save for patients hypersensitive to penicillin G, treatment with this drug offers both lower cost and risk than do alternative therapies. Although larger doses of penicillin may be given, for strains more resistant to penicillin, they suggest the consideration of alternative treatment. Among the latter are third-generation cephalosporins such as ceftriaxone and cefotaxime, although pneumococcal strains resistant to these ß-lactam drugs have also been recognized [33, 34]. Other agents of potential utility include vancomycin, macrolides (such as erythromycin and clindamycin), tetracyclines, cotrimoxazole, and chloramphenicol, although the last should be used rarely. If the cause of pneumonitis is unclear, the choice of antimicrobial agents must be based on clinical considerations of causes other than the pneumococcus.

Pneumococcal meningitis poses a special therapeutic problem because of the inability to achieve levels of penicillin in the subarachnoid space adequate to inhibit penicillin-resistant pneumococci [35]. For the treatment of meningitis caused by strains of pneumococci sensitive to third generation cephalosporins, these agents, such as ceftriaxone and cefotaxime, have proved useful, although, as noted, pneumococci may develop resistance to them. Because of the severity of pneumococcal meningitis and in the absence of known pneumococcal resistance to vancomycin, it seems prudent to include it in the initial treatment of all cases of pneumococcal meningitis in addition to a ß-lactam in areas in which intermediately or highly resistant pneumococci have been identified until the sensitivities of the infecting organism have been ascertained. Experience with vancomycin in the treatment of pneumococcal meningitis is limited, and some deaths have been reported when it has been used [36], but such an outcome is not unexpected when it is recalled that the case fatality rate among patients treated with penicillin for meningeal infection caused by penicillin-sensitive pneumococci may be as high as 20% to 40% among those older than 40 years. Further data are needed to define with more precision the optimal treatment of pneumococcal meningitis caused by penicillin-resistant or multidrug-resistant pneumococci.

In the face of the rising incidence of pneumococcal strains resistant to one or more antimicrobial agents, the simplest and potentially most effective means of dealing with the problems infections with them pose is to prevent their occurrence. Although the currently licensed polyvalent vaccine of pneumococcal capsular polysaccharides is recommended now only for persons older than 64 years or those with one of a variety of chronic systemic illnesses [37], there is no a priori reason to limit its administration to this segment of the population and not to immunize all adults against pneumococcal infection. Half the bacteremic infections currently identified in adults occur in individuals younger than 60 years, as was the case more than six decades ago. By relying on prophylaxis rather than on treatment, it should be possible to reduce significantly the incidence of severe and potentially lethal infections and the therapeutic dilemmas they impose when caused by a drug-resistant pathogen. A second potential advantage of widespread use of the vaccine is reduction of the pneumococcal carrier rate in the population. It has been known for half a century that the likelihood of becoming a carrier of a given pneumococcal type is reduced by approximately half after immunization, even though immunization will not eliminate the carrier state if it preceded vaccination [38]. Similar observations have been made after the administration of vaccines of other capsulated respiratory bacterial pathogens, including meningococci [39] and Haemophilus influenzae type b [40]. The impact of the use of the conjugate vaccine of H. influenzae type b capsular polysaccharide both on the incidence of infection and on the carrier rate has been dramatic. Protection of the highly vulnerable pediatric population against pneumococcal infection will have to await the further development of polyvalent polysaccharide-protein conjugate vaccines incorporating the capsular antigens of the pneumococcal types causing the preponderance of such illnesses in infancy and early childhood [41].